Mathematical English Usage

A Dictionary
by
Jerzy Trzeciak
Senior Copy Editor
Institute of Mathematics
Polish Academy of Sciences
Updated August 8, 2012
Available online at: http://www.impan.pl/Dictionary
c _ Copyright by Jerzy Trzeciak, Warszawa 2012
2 JERZY TRZECIAK
A
a, an We conclude that there is a smallest integer p for which f(p) = 0. Theorem 2 has a very
important converse, the Radon-Nikodym theorem. Our present assumption implies that the
last inequality in (8) must actually be an equality. Some of the isomorphisms classes above
will have a rank of 2. The only additional feature is the appearance of a factor of 2. This
says that f is no longer than the supremum of the boundary values of G, a statement similar
to (1). This term derives from quiver, a notion used in representation theory of algebras.
The earth has an average density 5.5 times that of water. If p = 0 then there are an additional
m arcs.
abbreviate [sth for sth; sth as sth; sth to sth; see also: brevity, short] We shall abbreviate the
expression (3) to F(k). We abbreviate this as f = g a.e. Thus, in abbreviated notation,.....
abbreviation [for sth; see also: shorthand, short] Note that (3) is merely an abbreviation for
the statement that.....
able [to do sth; see also: can] Using some facts about polynomial convexity, we are able to
deduce..... It seems plausible that..... but we have been able to establish this only in certain
cases. However, we have thus far been unable to nd any magic squares with seven square
entries.
abound [see also: numerous, plentiful, profusion, abundance, variety, wealth] Examples abound
in which P is discontinuous.
about [see also: roughly, approximately] The Taylor expansion of f about around) zero is.....
If s
0
lies below R
2
, then we can reect about the real axis and appeal to the case just considered.
These slits are located on circles about the origin of radii r
k
. The diameter of F is about
twice that of G. Then n(r) is about kr
n
. Let A denote the rectangle B rotated through /6 in
a clockwise direction about the vertex (0, 1). What would this imply about the original series?
What about the case where q > 2? It is hoped that a deeper understanding of these residues
will help establish new results about the distribution of modular symbols. On the other hand,
there is enormous ambiguity about the choice of M. In this section we ask about the extent
to which F is invertible. Here the interesting questions are not about individual examples,
but about the asymptotic behaviour of the set of examples as one or another of the invariants
(such as the genus) goes to innity. However, as we are about to see, this complication is easily
handled. This brings about the natural question of whether or not there is any topology on
the set of all possible itineraries.
above [see also: foregoing, precede, previous] The function F is bounded above below) by 1.
By the above,..... Let T be an isometric semigroup as above. In the notation above In the
above notation),.....
absence [see also: lack] The location of the zeros of a holomorphic function in a region is
subject to no restriction except the obvious one concerning the absence of limit points in .
[Note the dierence: absence = non-presence; lack = shortage of something desirable.]
absorb The second term can be absorbed by the rst.
abstract It seems that the relations between these concepts emerge most clearly when the setting
is quite abstract, and this (rather than a desire for mere generality) motivates our approach to
MATHEMATICAL ENGLISH USAGE 3
the subject. This abstract theory is not in any way more dicult than the special case of the
real line.
abundance [see also: wealth, variety, profusion, numerous] The monograph is illustrated with
an abundance of gures and diagrams.
abuse By abuse of notation, we continue to write f for f
1
. We shall, by convenient abuse
of notation, generally denote it by x
t
whatever probability space it is dened on. With the
customary abuse of notation, the same symbol is used for.....
accessible [to sb] Thus the paper is intended to be accessible both to logicians and to topologists.
The present paper is motivated by the desire to make the subject as accessible as possible.
accidental [see also: deliberately, intentionally] The use of the word generic is not accidental
here.
accomplish [see also: complete, nish, do, perform] We can accomplish both tasks by showing
that..... Actual construction of..... may be accomplished in a variety of ways.
accord [see also: agree, correspond, match] This accords with the intuition that as we pass down
the coding tree, we nd out more and more detailed information about the ordering actually
represented.
accordance [see also: agreement] Choose in accordance with Section 8.
according [as X or Y; to sth; to whether X or Y] The solutions are f or g according as t = 1 or
t = 2. Then F can be decomposed according to the eigenspaces of P. Choose S
k
according
to the following scheme. The middle part of Table 2 compares the classication according to
max a
i
, where only the longitudinal information is utilized, with those according to max b
ik
,
where both longitudinal and survival information are used.
accordingly [see also: respectively, suitably] The player has to decide which of the two strategies
is better for him and act accordingly. Write A = BC, factor a = bc accordingly, and let.....
account 1 [of sth; see also: description] A very readable account of the theory has been given in
[Zag]. We are indebted here to Villanis account (see [2]) of a standard generalization of convex
conjugacy. For a recent account we refer to [4]. See the simplied account in [2, Section 4].
See [17] for a brief account of the results obtained. [Not: the obtained results] The Markov
chain z
k
takes no account of how long the process stays in V . On account of (5), we have.....
2 [for sth; see also: explain, justify, reason, represent] This theorem accounts for the term
subharmonic. [= explains] So all the terms of (2) are accounted for, and the theorem
is proved. He accounts for all the major achievements in topology over the last few years.
[= He records] Firms employing over 1000 people accounted for 50% of total employment.
[= represented 50%]
accurate [see also: precise] The zeros of L-functions are all accurate to within 10
5
.
achieve [see also: attain, reach, take, gain] Equality is achieved only for a = 1. The function
g achieves its maximum at x = 5. Among all X with xed L
2
norm, the extremal properties
are achieved by multiples of U. This achieves our objective of describing.....
achievement [see also: result] He accounts for all the major achievements in topology over the
last few years. Their remarkable achievement seemed to validate Johns claim. However, it
4 JERZY TRZECIAK
soon turned out that..... a considerable extraordinary/ne/important/impressive/outstand-
ing/signicant) achievement
acknowledge We acknowledge a debt to the paper of Black [7]. This research was initiated when
the rst author was visiting the University of Alberta in the summer of 2008; the nancial support
and kind hospitality are gratefully acknowledged. We also acknowledge useful discussions with
J. Brown. The author gratefully acknowledges the referees helpful comments pertaining to the
rst draft of this paper. [Do not write: I acknowledge Dr. Brown for..... if you mean: I wish
to thank Dr. Brown for.....]
acquire [see also: get] Thus, the tensor algebra acquires a graded algebra structure.
across Along the way, we come across some perhaps unexpected rigidity properties of familiar
spaces.
act [see also: operate, proceed] We make G act trivially on Y . The player has to decide which
of the two strategies is better for him and act accordingly.
action All of the action in creating S
i+1
takes place in the individual cells of type 2 or 3. Away
from critical points, the action of G is reminiscent of the action of a cyclic group of order d.
actual Actual construction of..... may be accomplished in a variety of ways.
actually [see also: fact, more] The operator A is not merely symmetric, but actually selfadjoint.
Actually, Theorem 3 gives more, namely,..... Actually, the proof gives an even more precise
conclusion:..... Although the denition may seem articial, it is actually very much in the spirit
of Darbos old argument in [5]. Our present assumption implies that the last inequality in (8)
must actually be an equality. This accords with the intuition that as we pass down the coding
tree, we nd out more and more detailed information about the ordering actually represented.
adapt [,= adopt; see also: adjust, alter, change, convert, modify, vary, t, suitable] The proof
of Theorem 5 is easily adapted to any open set. The method of proof of Theorem B can be
adapted to extend the right-to-left direction of Mostowskis result by showing that..... This
denition is well adapted for dealing with meromorphic functions. The hypotheses of [4] are
dierent, however, and do not seem to adapt easily to the time-inhomogeneous case.
adaptation [see also: adjustment, modication, variation] Our method of proof will be an
adaptation of the reasoning used on pp. 7172 of [3].
add [see also: sum, total] Adding equations (2) and (3), we obtain..... [Not: Adding by sides]
The terms with n > N add up to less than 2.
addition Addition of (2) and (3) gives..... The set S is a semigroup with respect to coordi-
natewise addition. If h is modied by the addition of a suitable constant, it follows that.....
In addition to illustrating how our formulas work in practice, it provides a counterexample
to Browns conjecture. In addition to f being convex, we require that..... In addition to a
contribution to W
1
, there may also be one to W
2
. Assume, in addition to the hypotheses of
Exercise 4, that.....
additional [see also: extra, further, more] Now F has the additional property of being convex.
This solution has the additional advantage of being easily computable. If p = 0 then there
are an additional m arcs. [Note the article an.]
MATHEMATICAL ENGLISH USAGE 5
additionally [see also: also, moreover] Now F is additionally assumed to satisfy.....
address [see also: deal, take up] Strong compactness will be addressed in Section 3. The
main problems that we address are..... Addressing this issue requires using the convergence
properties of Fourier series. We close this article by addressing, in part, the case of what
happens if we replace the map T by convolution.
adhere [to sth; see also: adopt] We adhere to the convention that 0/0 = 0.
ad hoc [see also: provisional, temporary] With Lemma 2 in mind, we make the following ad
hoc denition.
adjoin If we adjoin a third congruence to F, say a = b, we obtain..... The extended real number
system is R with two symbols, and , adjoined.
adjust [see also: alter, adapt, change, modify] In the latter case we may simply adjust F to
equal 1 on the Borel set where it falls outside the specied interval. The constants are so
adjusted in (6) that (8) holds.
adjustment [see also: adaptation, modication] Although these proofs run along similar lines,
there are subtle adjustments necessary to t the argument to each new situation.
admit [sth; of sth] The continuum Y is tree-like since it admits a map onto X. This inequality
admits of a simple interpretation.
adopt [,= adapt; see also: adhere, take] We adopt throughout the convention that compact
spaces are Hausdor. We adopt the convention that the rst coordinate i increases as one goes
downwards, and the second coordinate j increases as one goes from left to right. To avoid
undue repetition in the statements of our theorems, we adopt the following convention. This
is the point of view adopted in Section 3. Furthermore, adopting this strategy considerably
eases constructing a coding tree from a linear order. We could have adopted an approach to
proving Theorem 2 along a line of reasoning which bears greater resemblance to the treatment
of the analogous result in Section 1.
advance [see also: development] The primary advance is to weaken the assumption that H is
C
2
, used by previous authors, to the natural condition that H is C
1
. Remarkable advances
have been made recently in the understanding of.....
advantage [of sth over sth; of doing sth; see also: merit, benet] One major advantage of..... is
that..... The advantage of using..... lies in the fact that..... This solution has the additional
advantage of being easily computable. This approach fails to take advantage of the Gelfand
topology on the character space. We take advantage of this fact on several occasions, by not
actually specifying the topology under consideration. On the other hand, as yet, we have
not taken advantage of the basic property enjoyed by S: it is a simplex. a considerable
decisive/denite/obvious/main/signicant) advantage
advantageous [see also: helpful, useful] In this case it is advantageous to transfer the problem
to (say) the upper half-plane.
advent However, with the recent advent of simulation based inference, the need for analytically
tractable posteriors is no longer critical.
6 JERZY TRZECIAK
aect [see also: inuence] Altering nitely many terms of the sequence u
n
does not aect the
validity of (9). We show that one can drop an important hypothesis of the saddle point theorem
without aecting the result. How is the result aected if we assume merely that f is bounded?
If a, b, and c are permuted cyclically, the left side of (2) is unaected. Properties involving
topological centres are unaected by a change to an equivalent weight.
armative We give an armative answer to the question of [3].
aord [see also: provide, furnish, yield] A counterexample is aorded by the KleinGordon
equation. We can now pose a problem whose solution will aord an illustration of how (5) can
be used. Having illustrated our method in Section 2, we can aord to be brief in our proof of
Theorem 5.
aeld To go into this in detail would take us too far aeld.
aforementioned Our rst result generalizes (8) by exploiting some general facts seemingly over-
looked by the aforementioned authors. We underline that the aforementioned results in [1] all
rely on the conformality of the underlying construction.
after The proof of (8) will be given after we have proved that..... We defer the proof of the
moreover statement in Theorem 5 until after the proof of the lemma. After making a linear
transformation, we can assume..... The desired conclusion follows after one divides by t and
lets t tend to 0. However, as pointed out right after (5),.....
again Hence, by (7) again, we have..... Finally, case (E) is completed by again invoking Theo-
rem 1. The operator H is again homogeneous.
against No specic evidence against the conjecture has been produced yet.
agree [see also: accord, correspond, match] Our denition agrees with the one of [3]. The
liftings on A and B agree on A B, hence we can piece them together to obtain.....
agreement [see also: accordance] This is in agreement with our previous notation. With this
agreement, it is clear that.....
aid [see also: help] The solutions can be carried back to H(V ) with the aid of the mapping
function . We see with the aid of an integration by parts that..... We now construct a group
that will be of aid in determining the order of G.
aim 1 [see also: desire, end, object, objective, task, purpose, intention] Our rst aim is to study
the ergodic properties of T. Our aim here is to give some sort of functorial description of K
in terms of G. This connection has been exploited to construct various innite families, with
the aim of lling possible gaps. In the remainder of this section, we study some properties
of K, with the eventual aim (not realized yet) of describing K directly using G. the broad
general/central/main/major/primary/limited/modest/underlying/original) aim
2 [to do sth, at sth, for sth; see also: design, intend] We aim to prove the following inequality:.....
These results therefore describe the very close connection between the method of encoding and
the structures we are aiming to classify. Aiming for a contradiction, suppose that.....
alas [see also: unfortunately] Having established (1), one might be tempted to try to extend
this result to general p through the choice of a suitable ideal B. Alas, as we shall see now, this
attempt is futile.
MATHEMATICAL ENGLISH USAGE 7
albeit [= though] However, we shall show in Section 3 that this simply results in Denition 3
again, albeit with complex weight. It is proved in [1] (albeit with a slightly dierent formulation)
that.....
algebra By elementary algebra, we can show that.....
algorithm It is obvious that the above theorem supplies an algorithm to eectively recognize
whether SP is in A. He used a new version of an algorithm for nding all normal subgroups
of up to a given index in a nitely presented group.
all [see also: any, each, every, whole, total] Hence all that we have to do is choose an x in X
such that..... Thus, all that remains is to repeat the construction for f in place of g. An
examination of the argument just given reveals that this is all we have used. All but a nite
number of the G
s
are empty. Note that any, but not all, of the sets h
1
and g
1
can be
empty. Now E, F and G all extend to U. a manifold all of whose geodesics are closed [= a
manifold whose geodesics are all closed] They all have their supports in V . They are all zero
at p. This map extends to all of M. These volumes bring together all of R. Bings published
mathematical papers. If t does not appear in P at all, we can jump forward n places. But
A
n
z
n
is much larger than the sum of all the other terms in the series

A
k
z
k
. Thus A is the
union of all the sets B
x
. the space of all continuous functions on X the all-one sequence
Any vector with three or fewer 1s in the last twelve places has at least eight 1s in all. The
elements of G, numbering 122 in all, range from 9 to 2000.
allow [sth; sth to be sth; sb to do sth; see also: enable, permit, possible] These theorems allow
one to guess the Plancherel formula. [Or: allow us to guess; not: allow to guess] As the
space of Example 3 shows, complete regularity of X is not enough to allow us to do that.
This allows proving the representation formula without having to integrate over X. This easily
allows the cases c = 1, 2, 4 to be solved. By allowing f to have both positive and negative
coecients, we obtain..... It is therefore natural to allow (5) to fail when x is not a continuity
point of F. The limit always exists (we allow it to take the value ). Lebesgue discovered
that a satisfactory theory of integration results if the sets E
i
are allowed to belong to a larger
class of subsets of the line. In [3] we only allowed weight functions that were C
1
. It should be
possible to enhance the above theorem further by allowing an arbitrary locally compact group L.
Here we allow a = 0. We deliberately allow that a given B may reappear in many dierent
branches of the tree.
allude [to sth; see also: mention, refer] We now come to the theorem which was alluded to in
the introduction of the present chapter. One should remark that the ambiguity alluded to in
Remark 3 disappears when talking about an ane eld.
almost [see also: nearly, practically] It is almost as easy to nd an element..... Incidentally,
the question of whether K(E) is amenable for specic Banach spaces E seems to have received
almost no attention in the literature.
alone [see also: only, solely] Neither (1) nor (2) alone is sucient for (3) to hold. Now M does
not consist of 0 alone. Then F is a function of x alone.
along This is derived in Section 3 along together) with a new proof of Morgans theorem.
The proof proceeds along the same lines as the proof of Theorem 5, but the details are more
complicated. For direct constructions along more classical lines, see [KL]. Although these
8 JERZY TRZECIAK
proofs run along similar lines, there are subtle adjustments necessary to t the argument to each
new situation. Along the way, we come across some perhaps unexpected rigidity properties of
familiar spaces.
already This has already been proved in Section 4. This idea is very little dierent from what
can already be found in [2]. We put b in R unless a is already in. In the physical context
already referred to, K is the density of..... [Note the double r in referred.] Inserting additional
edges destroys no edges that were already present.
also [see also: moreover, furthermore, likewise, too] Hence f
n
also converges to f. We shall
also leave to the reader the proof of (5). Since R is a polynomial in x, so also is P. The map
G is also not C
1
. It is also not dicult to obtain the complete additivity of . The k
i
are
not values of u and hence also not values of w.
alter [see also: adjust, change, convert, modify, transform, vary] We shall need ways of con-
structing new triangulations from old ones which alter the f-vector in a predictable fashion.
Altering nitely many terms of the sequence u
n
does not aect the validity of (9). The theorem
implies that some nite subcollection of the f
i
can be removed without altering the span.
alternate The terms of the series (1) decrease in absolute value and their signs alternate.
Successive vertices on a path have alternating labels.
alternately Every path on G passes through vertices of V and W alternately.
alternative An alternative way to analyze S is to note that..... Here is an alternative phrasing
of part (1):..... An ingenious alternative proof, shorter but still complicated, can be found
in [MR].
alternatively Alternatively, it is straightforward to show directly that.....
although [see also: though] Although [1] deals mainly with the unit disc, most proofs are
so constructed that they apply to more general situations. Although these proofs run along
similar lines, there are subtle adjustments necessary to t the argument to each new situation.
Although the denition may seem articial, it is actually very much in the spirit of Darbos old
argument in [5]. Now f is independent of the choice of (although the integral itself is not).
Thus, although we follow the general pattern of proof of Theorem A, we must also introduce new
ideas to deal with the lack of product structure. Although standard, the notion of a virtual
vector bundle is not particularly well known.
altogether [see also: completely, total, whole] However, we prefer to avoid this issue altogether
by neglecting the contribution of B to S. There are forty-three vertices altogether.
always The problem is that, whatever the choice of F, there is always another function f such
that..... The induced topology is not compact, but we can always get it to be contained in a
Bohr topology. The vector eld H always points towards the higher A-level.
ambiguity On the other hand, there is enormous ambiguity about the choice of M. One should
remark that the ambiguity alluded to in Remark 3 disappears when talking about an ane eld.
When there is no ambiguity we drop the dependence on B and write just Y
T
for Y
T,B
. This
also resolves the ambiguity introduced earlier in choosing an order of the lifts of U.
among [= amongst; see also: between, of, out of, include] Among the attempts made in this
direction, the most notable ones were due to Jordan and Borel. Among all X with xed L
2
MATHEMATICAL ENGLISH USAGE 9
norm, the extremal properties are achieved by multiples of U. If a
n
is the largest among
a
1
, ..., a
n
, then..... Our main results state in short that MEP characterizes type 2 spaces
among reexive Banach spaces. The existence of a large class of measures, among them that
of Lebesgue, will be established in Chapter 2. There are several theorems for a number of
other varieties. Among these are the Priestley duality theorem and..... the number of solutions
(x
1
, . . . , x
n
) in which there are fewer than r distinct values amongst the x
i
The next corollary
shows among other things that..... [Not: among others] Our result generalizes Urysohns
extension theorems, among others. [= among other theorems]
amount 1 The quantities F and G dier by an arbitrarily small amount. Thus will be less
than by an amount comparable to a(s). It is intuitively clear that the amount by which S
n
exceeds zero should follow the exponential distribution. It contributes half of the amount on
the right hand side of (1).
2 [to sth; to doing sth; see also: total, add up] When n = 0, (7) just amounts to saying
that..... This just amounts to a choice of units. Consumer spending on those items amounts
to $9 billion. Internet sales still amounted to only 3% of all retail sales in November.
analogous [to sth] The theory of..... is entirely completely) analogous to..... We shall also
refer to a point as backward nonsingular, with the obvious analogous meaning. Using (2) and
following steps analogous to those above, we obtain.....
analogously [to sth] The lower limit is dened analogously: simply interchange sup and inf
in (1). The notion of backward complete is dened analogously by exchanging the roles of f
and f
1
. Analogously to Theorem 2, we may also characterize.....
analogue [of sth; amer. analog] This is an exact analogue of Theorem 1 for closed maps. No
analogue of such a metric appears to be available for Z.
analogy Let us see what such a formula might look like, by analogy with Fourier series. In
analogy with (1) we have..... There is a close analogy between..... The analogy with statistical
mechanics would suggest..... Our presentation is therefore organized in such a way that the
analogies between the concepts of topological space and continuous function, on the one hand,
and of measurable space and measurable function, on the other, are strongly emphasized.
analyse A model for analysing rank data obtained from several observers is proposed.
analysis [see also: exploration, investigation, study] The only case requiring further analysis
occurs when f = 0. We now transfer the above analysis back to M(A). The analysis is
similar to that of [3]. Analysis of the proofs of these previous results shows that..... These
results show that an analysis purely at the level of functions cannot be useful for describing.....
a careful close/comprehensive/detailed/systematic/thorough) analysis
angle Draw the half-line from x at angle to . The line T makes a right angle with the chosen
direction. Then F and G make angle . Let ABC be an angle of sixty degrees. The two
lines intersect at an angle of ninety degrees. As the point z moves around the unit circle, the
corresponding J
z
s are rotations of angle t(z).
announce Kim announces that (by a tedious proof) the upper bound can be reduced to 10.
anomalous The prime 2 is anomalous in this respect, in that the only edge from 2 passes
through 3.
10 JERZY TRZECIAK
another Another group of importance in physics is SL
2
(R). In the next section we introduce
yet another formulation of the problem. It has some basic properties in common with another
most important class of functions, namely, the continuous ones. We have thus found another
three solutions of (5). [= three more] Here the interesting questions are not about individual
examples, but about the asymptotic behaviour of the set of examples as one or another of the
invariants (such as the genus) goes to innity. Then one Y
i
can intersect another only in one
point. It is highly likely that if one of the Xs is exchanged for another, the inequality fails.
answer [see also: explanation, solution] An armative answer is given to the question of [3]
whether..... When A is commutative, the answer to both questions is yes. The algorithm
returns 0 as its answer. The answer depends on how broadly or narrowly the term matrix
method is dened. The answer is not known to us. What is the answer if a = 0?
any [see also: arbitrary, all, each, every, whatever, whichever] By deleting the intervals con-
taining x, if any, we obtain..... There are few, if any, other signicant classes of processes for
which such precise information is available. Let Q denote the set of positive denite forms
(including imprimitive ones, if there are any). The preceding denitions can of course equally
well be made with any eld whatsoever in place of the complex eld. If K is now any compact
subset of H, then there exists..... Note that F(t) may only be dened a.e.; choose any one
determination in (7). Note that any, but not all, of the sets h
1
and g
1
can be empty.
for any two triples [Not: for every two triples; every requires a singular noun.]
apart [from sth; see also: besides, except, distinguish] Apart from these two lemmas, we make
no use of the results of [4]. That isapart from the use of relaxed controlsprecisely the
stochastic Bellman equation. Apart from being very involved, the proof requires the use of.....
There is a curve lying entirely in the open strip 0 < < 1 apart from the endpoints such
that..... for no x apart from the unique solution of..... [Note the dierence between besides,
except and apart from: besides usually indicates adding something, except subtracts, and
apart from can be used in both senses; after no, nothing etc., all three can be used.] Their
centres are a distance at least N apart. The m points x
1
, . . . , x
m
are regularly spaced t units
apart. What sets the case n = 5 apart is the fact that homotopic embeddings in a 5-manifold
need not be isotopic.
apparatus Keller, in his fundamental paper [7] concerning duality, develops an apparatus that
allows him to obtain a very wide variety of duality theorems.
apparent [see also: clear, evident, obvious, plain] If one studies the proof of..... it is apparent
that (2) is never used. It is now apparent what the solution for K will be like:..... This is
usually called the area theorem, for reasons that will become apparent in the proof.
apparently [see also: seemingly] Fox has apparently [= as one can see] overlooked the case
of..... Note that the apparently [= seemingly] innite product in the denominator is in fact
nite. The reader may wonder why we have apparently ignored the possibility of obtaining a
better lower bound by considering.....
appeal 1 [see also: recourse] Recently proofs have been constructed which make no appeal to
integration. In the preceding proof, the appeal to the dominated convergence theorem may
seem to be illegitimate since..... Through an appeal to (5.3) we have.....
2 [see also: invoke, refer] At this stage we appeal to Theorem 2 to deduce that..... We can
also appeal to Lemma 5 to see that the uniform continuity condition (5.3) is met. If s
0
lies
MATHEMATICAL ENGLISH USAGE 11
below R
2
, then we can reect about the real axis and appeal to the case just considered. One
of the appealing aspects of the spectral set is that it readily lends itself to explicit computation.
appear [see also: look, occur, seem, turn out, turn up, look] However, no extension in this
direction has appeared in the literature. The statement does appear in [3] but there is a simple
gap in the sketch of proof supplied. It will eventually appear that the results are much more
satisfactory than one might expect. The zeros appear at intervals of 2m. Every prime in the
factorization appears to an even power. No analogue of such a metric appears to be available
for Z. Conditions relating to bounds on the eigenvalues appear to be rare in the literature.
At rst glance, this appears to be a strange denition. [= seems to be] Neighbourhoods of
points in these spaces appear at rst glance to have a nice regular structure, but upon closer
scrutiny, one sees that many neighbourhoods contain collections of arcs hopelessly folded up.
This may appear rather wasteful, especially when n is close to m, but these terms only give a
small contribution to our sum. It does not appear feasible to adapt the methods of this paper
to.....
appearance The only additional feature is the appearance of a factor of 2. This convention
simplies the appearance of results such as the inversion formula.
applicability The abstract theory gives us a tool of much wider applicability.
applicable [to sth] We now provide a bound applicable to systems of..... The hypothesis n > 1
ensures that Lemma 2 is applicable.
application [see also: means, use, via] Repeated application of (4) shows that..... As an
example of the application of Theorem 5, suppose..... Even in the case n = 2, the application
of Theorem 6 gives essentially nothing better than the inequality..... Specically, one might
hope that a clever application of something like Choquets theorem would yield the desired
conclusion. A drawback to Polya trees, and perhaps the main reason why they have not seen
much application within the Bayesian nonparametric literature, is that an arbitrary partition
tree needs to be specied. As an application, consider the Dirichlet problem Lf = 0. We
then show how this leads to stronger results in applications. [Do not use application when
you mean map: a map f : X Y (not: an application f).]
apply [to sth; sth to sth; sth to do sth; see also: employ, use, utilize, pertinent, relevant] Now (5)
follows from (4) if (2) is applied to the last equation. We now apply the previous observation
to estimate F. We apply this to g to obtain..... In particular, the theorem applies to weakly
conuent maps. We nish by mentioning that, suitably modied, the results of Section 2 apply
to the AP case. More generally this argument also applies to characterizing Hurewicz subsets
of I. Then the same argument as in Theorem 5 applies to show that L(R) fails to be amenable.
Actually, [3, Theorem 2] does not apply exactly as stated, but its proof does.
appreciation He would like to express his appreciation to the faculty and sta of the Dartmouth
mathematics department for their hospitality.
approach 1 [to sth; to doing sth; see also: method, technique, procedure, way, line] We take
the same approach as in [3]. We could have adopted an approach to proving Theorem 2 along
a line of reasoning which bears greater resemblance to the treatment of the analogous result in
Section 1. Very recently, Heck introduced a general approach that unies and extends all these
results. The case a = 1 requires a dierent approach. It seems that the relations between
12 JERZY TRZECIAK
these concepts emerge most clearly when the setting is quite abstract, and this (rather than a
desire for mere generality) motivates our approach to the subject. We sketch below one possible
approach to obtaining such renements. The novelty of our approach lies in using..... It is
important to notice some of the weaknesses inherent in the above approach The approach in
[GT] provides a unied way of treating a wide variety of seemingly disparate examples.
2 [see also: converge, tend] In the study of innite series

a
n
it is of signicance whether the
a
n
approach zero rapidly.
appropriate [see also: suitable, convenient, good, t, correspond] Theorems 3 and 6 of [2],
with the appropriate changes, are also valid. By writing out the appropriate equations, we see
that this is equivalent to..... Perhaps it is appropriate at this point to note that a representing
measure is countably additive if and only if..... It seems appropriate to mention in passing the
corresponding formula for the cohomology of B
n
.
appropriately [see also: conveniently, suitably] Now choose t appropriately as a function of .
If A is such an operator (appropriately chosen) then..... Suppose that S is an (appropriately
normalized) positive current on M.
approximate 1 an approximate solution
2 It is therefore reasonable that the behaviour of p should in some rough sense approximate the
behaviour of q. Here Y is a Poisson variable suitably chosen to approximate X in distribution.
a sequence of smooth domains that approximates D from within
approximately [see also: roughly, about] Then x R
j
for some j which is approximately equal
to m+k. When we pass from T to T +, f(z) changes by approximately M. in the ratio of
approximately 3 : 1
arbitrarily The quantities F and G dier by an arbitrarily small amount. [Not: arbitrary
small] .....where C can be made arbitrarily small by taking..... Runges theorem will now be
used to prove that meromorphic functions can be constructed with arbitrarily preassigned poles.
arbitrary [see also: all, each, every, whatever, whichever] This enables us to dene solution
trajectories x(t) for arbitrary t. The theorem indicates that arbitrary multipliers are much
harder to handle than those in M(A). One cannot in general let A be an arbitrary substructure
of B here. If X happens to be complete, we can dene f on E in a perfectly arbitrary manner.
area [see also: eld] The region A has an area of 15 m
2
. This is an interesting area for future
research. This is an area where there is currently a lot of activity.
argue [see also: assert, claim, reason] To see that A = B we argue as follows. But if we argue
as in (5), we run into the integral....., which is meaningless as it stands. Arguing by duality we
obtain..... It might be argued that the h-principle gives the most natural approach to.....
argument [see also: reasoning] A similar argument holds for the other cases. A deformation
retract argument completes the proof. In outline, the argument follows that of the single-valued
setting, but there are several signicant issues that must be addressed in the n-valued case.
This argument comes from [4]. This argument is invalid for several reasons. However, this
argument is fallacious, because as remarked after Lemma 3,..... By an elementary argument,.....
This is handled by a direct case-by-case argument. Following the argument in [3], set.....
The case f = 1 requires a dierent argument. But the T
n
need not be contractions in L
1
, which
is the main obstruction to applying standard arguments for densities. We give the argument
MATHEMATICAL ENGLISH USAGE 13
when I = R. Continuity then nishes o the argument. This completes our argument for (1).
This assumption enables us to push through the same arguments. It simplies the argument,
and causes no loss of generality, to assume.....
arise [see also: emerge, occur, result] ....., the last equality arising from (8). This case arises
when..... The question arises whether..... A further complication arises from BP, which
works rather dierently from the other labels. If no confusion can arise, we write K for both
the operator and its kernel.
around The Laurent expansion of f around about) zero is..... To get around this delicate issue,
we shall separate the variables s and u. To get around overcome) this diculty, assume.....
As the point z moves around the unit circle, the corresponding J
z
s are rotations of angle t(z).
arrange [see also: order, organize] We wish to arrange that f be as smooth as possible. We
can arrange the X
i
s to include every element of F. Let (a
n
) be the sequence of zeros of f
arranged so that [a
1
[ [a
2
[ ..... The present proof is so arranged that it applies without
change to holomorphic functions of several variables.
arrangement [see also: order, organization] Since most of the results presented are quite clas-
sical (the novelty lies in the arrangement, and some of the proofs are new), I have not attempted
to document the source of every item.
arrive [at sth; see also: reach] Setting f = 0, we arrive at a contradiction.
article [see also: paper] We close this article by addressing, in part, the case of what happens
if we replace the map T by convolution. We can now formulate the problem to which the rest
of this article is dedicated.
articial Although the denition may seem articial, it is actually very much in the spirit of
Darbos old argument in [5].
as [see also: like, since] If f is as in (8), then..... [Not: like in (8)] We can multiply two
elements of E by concatenating paths, much as in the denition of the fundamental group.
....., where each function g is as specied described) above. Actually, [3, Theorem 2] does not
apply exactly as stated, but its proof does. They were dened directly by Lax [2], essentially as
we have dened them. For k = 2 the count remains as is. In the case where A is commutative,
as it will be in most of this paper, we have..... As a rst step we identify the image of .
Then F has T as its natural boundary. The algorithm returns 0 as its answer. Now X can
be taken as coordinate variable on M. If one thinks of x, y as space variables and of z as time,
then..... Then G is a group with composition as group operation. We have A B as right
modules. Then E is irreducible as an L-module. ....., as is easily veried. ....., as noted as
was noted) in Section 2. [Not: as it was noted] ....., as desired claimed/required). The
elements of F are not in S, as they are in the proof of..... Note that F is only nonnegative
rather than strictly positive, as one may have expected. Then G has 10 normal subgroups and
as many non-normal ones. Moreover, H is a free R-module on as many generators as there are
path components of X. But A has three times as many elements as B has. We can assume
that p is as close to q as is necessary for the following proof to work. Then F can be as great
as 16. Each tree is about two-thirds as deep as it was before. As M is ordered, we have no
diculty in assigning a meaning to (a, b). The ordered pair (a, b) can be chosen in 16 ways so
as not to be a multiple of (c, d). Now (3) is clear. As for (4), it is an immediate consequence
14 JERZY TRZECIAK
of Lemma 6. [= Concerning (4)] As with the digit sums, we can use alternating digit sums to
prove..... [= Just as in the case of digit sums]
ask In this section we ask about the extent to which F is invertible. This is the same as asking
which row vectors in R have diering entries at positions i and j. An obvious question to ask
is whether the assertion of Theorem 1 continues to hold for.....
aspect [see also: detail, feature, characteristic, ingredient, point] One of the appealing aspects
of the spectral set is that it readily lends itself to explicit computation. I shall limit myself
to three aspects of the subject. We shall touch only a few aspects of the theory. a central
crucial/fundamental/key/main/major/principal/fascinating/striking/neglected) aspect
assert [see also: say, state] The spectral radius formula asserts that..... Puiseuxs theorem
asserts the existence of..... We also need the following technical lemma, which asserts the
rarity of numbers with an inordinately large number of prime factors. Here is a more explicit
statement of what the theorem asserts. To prove the asserted convergence result, rst note
that.....
assertion [see also: conclusion, statement] Now (2) is clearly equivalent to the assertion that.....
....., which proves the assertion. [Not: the thesis] If we prove (8), the assertion follows.
The interest of the lemma is in the assertion that..... Assertion (b) is known as the Radon-
Nikodym theorem.
assess [see also: estimate] To assess the quality of this lower bound, we consider the following
special case.
assign [see also: associate] The map f assigns to each x the unique solution of..... As M is
ordered, we have no diculty in assigning a meaning to (a, b). A weighted graph is one in
which each vertex is assigned an integer (called its weight). Here the variable h is assigned
degree 1.
assignment The assignment of K
1
to K, and of T
1
to T, denes a functor between the category of
commutative algebras and the category of compact semigroups with continuous homomorphisms.
associate [sth with/to sth; see also: assign, connect, link, relate, join] With each D there is
associated a region V
D
. Associated with each Steiner system is its automorphism group, that
is, the set of all..... Let l be an eigenvalue of A, and let v be an associated eigenvector.
Generally we add a tilde to distinguish between quantities associated to

G and those associated
to G.
assume [see also: suppose, presume] Here F is assumed to be open. [But: We assume that F is
open, not: We assume F to be open.] ....., the limit being assumed to exist for every real x.
The assumed positivity of u
n
is essential for these results. We can assume, by decreasing
n if necessary, that..... We may (and do) assume that..... We tacitly assume that..... It is
assumed that..... We follow Kato [3] in assuming that f is upper semicontinuous. The reader
is assumed to be familiar with elementary K-theory. Then X assumes values 0, 1, ..., 9, each
with probability 1/10.
assumption [see also: condition, hypothesis, requirement] We make two standing assump-
tions on the maps under consideration. By the smoothness assumption on f,..... By as-
sumption,..... If the boundary is never hit then x
t
is a Feller process under reasonable con-
MATHEMATICAL ENGLISH USAGE 15
tinuity assumptions. We establish our results both unconditionally and on the assumption
of the Riemann Hypothesis. We note that the assumption of GCH is made for convenience
and ease of presentation. Then F is continuous at zero, contrary to assumption. a basic
fundamental/implicit/tacit/underlying/reasonable/erroneous) assumption
at At the fourth comparison we have a mismatch. At the suggestion of the referee, we consider
some simple cases. The match occurs at position 7 in T. Now R is the localization of Q at a
maximal ideal. Next, F preserves angles at each point of U. We may assume that this is the
rst point at which these two curves have met. at the end of Section 2 Now F is dened to
make G and H match up at the left end of I. The zeros appear at intervals of 2m. We can
make g Lipschitz at the price of weakening condition (i). The two lines intersect at an angle
of ninety degrees.
attach The name of Harald Bohr is attached to bG in recognition of his work on almost periodic
functions.
attain [see also: achieve, reach, take] Equality is attained only for a = 1. The function g
attains its maximum at x = 5. Now (c) asserts only that the overall maximum of f on U is
attained at some point of the boundary.
attempt 1 [at sth; at doing sth; to do sth; see also: trial] This work was intended as an
attempt to motivate at motivating)..... There are other problems with this example which
would hinder any attempt to follow the proof given here too closely. Among the attempts
made in this direction, the most notable ones were due to Jordan and Borel. This attempt is
doomed because the homogeneity condition fails to hold. [= The attempt is certain to fail]
Having established (1), one might be tempted to try to extend this result to general p through
the choice of a suitable ideal B. Alas, as we shall see now, this attempt is futile. a successful
unsuccessful/failed/serious/repeated) attempt
2 [sth; to do sth; see also: try] In 1988, while attempting to generalize this result, the second
author noticed that..... Since most of the results presented are quite classical (the novelty lies
in the arrangement, and some of the proofs are new), I have not attempted to document the
source of every item. Thus it is reasonable to attempt, using this homeomorphism, to gain an
understanding of the structure of M.
attention It is a pleasure to thank R. Greenberg for bringing his criterion for..... to our attention,
and for generously sharing his ideas about it. We now turn our attention to..... We can do
this by restricting attention to..... In this section I shall focus attention on..... From now on
we conne attention to R
2
. Incidentally, the question of whether K(E) is amenable for specic
Banach spaces E seems to have received almost no attention in the literature.
author The author thanks H. Miller for a careful reading of an earlier draft. This observation
prompted the author to look for a more constructive solution. The above construction has led
the author to believe that..... The authors interest in this problem was recently rekindled by a
conversation with David Lees. To the best of the authors knowledge, the problem is still open.
The only references known to the authors are [A] and [V], where the case A = L(E) is settled
in the negative. In 1988, while attempting to generalize this result, the second author noticed
that..... The paper was commenced whilst the second author held a Fullbright Fellowship.
Our rst result generalizes (8) by exploiting some general facts seemingly overlooked by the
aforementioned authors. This topic has been dealt with by many authors. A number of
16 JERZY TRZECIAK
authors have considered, in varying degrees of generality, the problem of determining.....
automatic An i-dependent lower bound is automatic by continuity.
automatically Note that f = limf
n
automatically exists.
auxiliary In Section 2 the reader will be reminded of some important properties of Bernoulli
numbers, and some auxiliary results will be quoted or derived.
available No analogue of such a metric appears to be available for Z. A further tool available
is the following classical result of Chen.
average 1 The observed values of X will on average cluster around points where..... How many
multiplications are done on average?
2 P rides a cycle at an average speed of 20 km an hour. The earth has an average density 5.5
times that of water.
3 The measure is then obtained by averaging the standard measures over all choices of.....
avoid [sth; doing sth; see also: miss] We should avoid using (2) here, since..... [Not: avoid
to use] However, we prefer to avoid this issue altogether by neglecting the contribution of
B to S. To avoid undue repetition in the statements of our theorems, we adopt the following
convention. They dened the concept solely in terms of the norm of the Banach space, delib-
erately avoiding any extra structure. This example indicates that the inductive argument in
the proof of Theorem 1 is unavoidable.
await Nevertheless we still await a full characterization of the Banach spaces E such that K(E)
is amenable.
aware [see also: know] At the time of writing [5], I was not aware of this reference. One must
also be aware that the curvature of M
i
might not be bounded uniformly in i. In 1925 Franklin,
unaware of Stackels work, showed.....
away [see also: beyond, o, outside, distance] Then F is smooth away bounded away) from
zero. Away from critical points, the action of G is reminiscent of the action of a cyclic group
of order d. It follows that z is at least 2MN away from the left endpoint of I.
B
back A linear transformation brings takes) us back to the case in which..... The solutions can
be carried back to H(V ) with the aid of the mapping function . .....(a result that dates back
to a 1915 paper of Hadamard) This idea goes back at least as far as [3]. This argument goes
back to Banach. Schenzels formula frequently allows us to move back and forth between the
commutative algebra of k[P] and the combinatorics of P. We now transfer the above analysis
back to M(A). We now turn back to our main question.
background For background information, see [5]. For relevant background material concerning
random walks, see [2]. Section 2 contains an overview of the necessary background.
backward(s) We shall also refer to a point as backward nonsingular, with the obvious analogous
meaning. The notion of backward complete is dened analogously by exchanging the roles of
f and f
1
. Extend this sequence of numbers backwards, dening N
1
, N
2
and N
3
by.....
Then G is simply g with its periodic string read backwards. Let A

be A run backwards.
MATHEMATICAL ENGLISH USAGE 17
[In most adverbial uses, backward and backwards are interchangeable; as an adjective, the more
standard form is backward: a backward shift.]
ball a ball of radius r about the origin the ball of radius r centred at x with centre at x)
base 1 [see also: basis] the base p expansion representation) of x From now on all logarithms
are to base two. Take as base for a topology on X the sets of the form..... For the base step
of the induction, consider a vertex t in A.
2 [see also: depend, rest, rely] Our proof of Theorem 2 is based upon ideas found in [BN]. We
base our development on two properties of prolongation peculiar to this case. For our purposes
here, the best way is to base the proof on the following theorem, a derivation of which can be
found in [P].
basic [see also: crucial, essential, critical, main, key, primary, elementary] On the other hand,
as yet, we have not taken advantage of the basic property enjoyed by S: it is a simplex. We
presume a basic knowledge of large cardinals and forcing. For the basics on tensor products,
see [Da].
basically [see also: essentially, mainly] As noted in the introduction, this is basically combining
Sawyers result with a variation of the arguments of Hunt.
basis [of sth, for sth; pl. bases; see also: base, foundation, underlie] Theorem 2 will form the
basis for our subsequent results. The basis of most of these theorems is Jensens formula.
be [see also: being] The theorem to be proved is the following. [= which will be proved] We
conclude with two simple lemmas to be used mainly in..... One can, for example, take A to
be the rationals in X. This assumption is certainly necessary if the distribution of x
t
is to
converge to F. If there are to be any nontrivial solutions x then any odd prime must satisfy.....
However, as we are about to see, this complication is easily handled.
bear An example to bear in mind is behaviour in the basin of a periodic point. There is a
fourth notion of phantom map which bears the same relation to the third denition as the rst
does to the second. We could have adopted an approach to proving Theorem 2 along a line
of reasoning which bears greater resemblance to the analogous result in Section 1. All three
cases bear a striking resemblance. This shape bears a striking similarity to that of..... While
we suspect that the methods of the present section can be brought to bear on this question also
[= be relevant to this question], we will not pursue this analysis in detail here.
because [see also: as, for, since] However, this argument is fallacious, because as remarked
after Lemma 3, K is discontinuous. This is because the factor M satises condition (P).
Unfortunately, because of the possible presence of cusps, this need not be true. Because of
this, W is never long enough to cancel with M in the product ABC.
become Then F becomes inner when extended to B. The conclusion of Theorem 3 becomes
false if this requirement is omitted. When h is in H, the integral formula becomes Af =.....
It becomes impracticable to compute the zeros of F for degrees greater than 6; in any event,
deciding whether the divisors found in this way represent irreducible curves becomes increasingly
dicult. Indeed, as n increases, it becomes increasingly rare for a manifold to be a hyperplane
section of another projective manifold.
before The sum of the depths is at most two-thirds of what it was before. Each tree is about
18 JERZY TRZECIAK
two-thirds as deep as it was before. The remainder of the estimation is largely as before with
B replaced by C. The proof concludes as before. Most of the theorems presented here have
never been published before. As noted before, there exists N homeomorphic to P such that.....
Note that P comes before Q along the arc l. We may require that the point P lie in one of
the trees constructed before or during the ith stage of the induction. [Note the subjunctive lie.]
beforehand For these considerations it will be convenient to state beforehand, for easy reference,
the following variant of.....
begin [with sth; by doing sth; see also: proceed, start, commence] We begin by describing the
class of functions f considered, which includes the special cases quoted above. We begin with
a detailed analysis of..... There is quite an extensive literature concerning resonance problems,
beginning with the work of Lazer. To begin, let A
0
be the trivial algebra with automorphism g
0
.
beginning [see also: initially, originally, rst, start, outset] (see the denition at the beginning
of Section 2) In the beginning [= Initially], he suspected that.....
behave Note that C behaves covariantly with respect to maps of both X and G. We need to
check that F-derivatives behave in the way we expect with regard to sums, scalar multiples and
products. Hence we would expect the functions F
i
to behave similarly. For general rings,
Out(R) is not necessarily well-behaved.
behaviour [amer. behavior] It is therefore reasonable that the behaviour of p should in some
rough sense approximate the behaviour of q. Some restrictions must be placed on the behaviour
of f. This gives some control over the behaviour of..... A function exhibiting this type
of behaviour has been constructed by..... We also show the existence of E-transformations
exhibiting nearly the full range of behaviours possible for scaling transformations.
behind The idea behind our use of the map is that..... This is the motivation behind the
following process.
being Note that M being cyclic implies F is cyclic. The probability of X being rational equals
1/2. This is exactly our denition of a weight being regulated. We have to show that the
property of there being x and y such that x < y uniquely determines P up to isomorphism.
In addition to f being convex, we require that..... Here J is dened to equal Af, the function
f being as in (3). [= where the function f is.....; not: being the function f as in] ....., the
constant C being independent of..... The ideal is dened by m =....., it being understood
that..... But....., it being impossible to make A and B intersect. [= since it is impossible
to make] The map F being continuous, we can assume that..... Actually, S has the much
stronger property of being convex. This method has the disadvantage of not being intrinsic.
[Do not write: the function f being the solution of (1) if you mean: the function f that is the
solution of (1).]
believe It is this proposition that we believe to be false in Morava E-theory. While nonpara-
metric priors are typically dicult to manipulate, we believe the contrary is true for quantile
pyramids. There is no reason to expect this to be an inverse map on K, but we do have the
following.
belong [see also: element, member, membership] Lebesgue discovered that a satisfactory theory
of integration results if the sets E
i
are allowed to belong to a larger class of subsets of the line.
Two consecutive elements do not belong both to A or both to B. It turns out that A, B and C
MATHEMATICAL ENGLISH USAGE 19
all belong to the same class, which we represent by the symbol P
2
. For the sake of clarity, we
shall indicate in what follows to which space X belongs. [Do not write: E belongs to the most
important classes of algebras if you mean: E is one of the most important classes of algebras
or E is among the most important classes of algebras.]
below A brief sketch of the reasoning is given below. By Remark 3 below,..... The pressure
increases are signicantly below those in Table 2. This proves that the dimension of S does not
go below q. a function bounded below above) by 1 As a rst step we shall bound A below.
benet [see also: advantage, merit] 1 The benet of formulating our notion of isomorphism sec-
tion as above will become clear shortly. For the benet of Pascal programmers, we explain.....
Although our proof is a little tedious, it is much less so than Itos original proof, which was
carried out without the benet of martingale technology.
2 In doing this we will also benet from having the following notation. We have beneted
greatly from discussions with many people, including.....
besides [see also: apart from, except] Besides being very involved, the proof requires the use.....
for no x besides the unique solution of..... [Note the dierence between besides, except and
apart from: besides usually adds something, except subtracts, and apart from can be used in
both senses; after no, nothing etc., all three can be used.] [Do not write: Besides, the solution
is given by (5) if you mean: Moreover, the solution is given by (5).]
best It is best not to use..... The best one could hope for is K F. This result is best
possible. [Or: the best possible] In Theorem 2 we show that the bound is not far from best
possible. The above bound on a
n
is close to best possible. The best known of these is the
Knaster continuum. It can be inferred from known results that these series at best converge
conditionally in L
p
.
better [see also: improve, improvement] Recent improvements in the HL-method enable us to
do better than this. In fact, we can do even better, and prescribe nitely many derivatives at
each point of A. Even in the case n = 2, the application of Theorem 6 gives essentially nothing
better than the inequality.....
between [see also: among] There cannot be two edges between one pair of vertices. Generally
we add a tilde to distinguish between quantities associated with

G and those associated with G.
It seems that the relations between these concepts emerge most clearly when the setting is
quite abstract, and this (rather than a desire for mere generality) motivates our approach to the
subject. The assignment of K
1
to K, and of T
1
to T, denes a functor between the category of
commutative algebras and the category of compact semigroups with continuous homomorphisms.
beware [see also: caution] Beware that the pro-objects of [Me] are indexed by categories that
are not necessarily small. Beware that more than one equation is referred to as the fast diusion
equation in the literature.
beyond [see also: away, o, outside] We would like to know....., but that is beyond our reach at
this point. It is beyond the scope of this paper to give a complete treatment of..... Ination
in the rst quarter rose beyond the acceptable level of 5%.
body We refer the reader to the body of the paper for details.
20 JERZY TRZECIAK
borrow We summarize some of its main properties, borrowing from the elegant discussion in
Hensons article.
both [see also: either] Both f and g are obtained by..... [Or: f and g are both obtained]
For both C

and analytic categories,..... Here C behaves covariantly with respect to maps

of both X and G. We now apply (3) to both sides of (6). Both conditions Both these
conditions/Both the conditions) are restrictions only on..... [Note: the after both.] Then C lies
on no segment both of whose endpoints lie in K. Two consecutive elements do not belong both
to A or both to B. Both its sides are convex. [Or: Its sides are both convex.] Here B and C
are nonnegative numbers, not both 0. Choose points x in M and y in N, both close to z, such
that..... We show how this method works in two cases. In both, C is.....
bottom [see also: low, top] Number the successive segments of the boundary line between A
and B (marked thickly in the picture) with the numbers 0, 1, ..., n, starting at the bottom. The
matrix A has an n m block of zeros in the bottom left-hand corner.
bound 1 [see also: estimate] The above bound on a
n
is close to best possible to the best
possible). We conclude that, no matter what the class of b is, we have an upper bound on M.
The main new feature is the use of the face ring to produce lower bounds for the number of
vertices. Kim announces that (by a tedious proof) the upper bound can be reduced to 10.
2 [see also: dominate, estimate] F is bounded above below) by a constant times f(z). Then
G is bounded away from zero. Furthermore, K is an upper bound on for) f(x) for x in K.
Theorem 1 can be used to bound the number of such T.
boundary Then F has T as its natural boundary.
bracket the expression in brackets in braces, in parentheses) the rst bracketed term on the
right of (6)
break [see also: decompose, split] The remainder of our work breaks into ve steps. In this
case the method of [4] breaks down. Then the sequence (8) breaks o in split exact sequences.
Thus the long exact sequence breaks up into short sequences.
brevity [see also: short, shorthand] For brevity, we drop the subscript t on h
t
. [Not: For
shortness]
brief [see also: concise, succinct] We conclude with a brief mention of free inverse semigroups.
We now turn to a brief discussion of another concept which is relevant to Johns theorem.
In brief outline, here is the main idea of the proof. We shall be brief here and just estimate
the terms that are the most challenging. Having illustrated our method in Section 2, we can
aord to be brief in our proof of Theorem 5.
briey [,= shortly] We denote it briey by c. [Not: shortly] We briey outline this renement
at the end of Section 2. [Not: in the end] The proof will only be indicated briey. There
has since been a series of improvements, of which we briey mention the work of Levinson.
bring We must now bring dependence on d into the arguments of [9]. The map F can be
brought into this form by setting..... Brucks theorem on common xed points for commuting
nonexpansive mappings is then brought into play by noting that..... Mary Lane deserves our
special thanks for her part in bringing this volume to a successful completion. It is a pleasure
to thank R. Greenberg for bringing his criterion for..... to our attention, and for generously
MATHEMATICAL ENGLISH USAGE 21
sharing his ideas about it. This brings about the natural question of whether or not there is
any topology on the set of all possible itineraries. A linear transformation brings us back to the
case in which..... Later, by bringing in the injectivity radius, Fong simplied the argument.
The aim of this paper is to bring together two areas in which..... These volumes bring together
all of R. Bings published mathematical papers.
broad [see also: wide] Most of these phenomena can be subsumed under two broad categories.
[= included in] Those results permit us to view this fragment of the theory from a broader
perspective.
broadly [see also: largely] The answer depends on how broadly or narrowly the term matrix
method is dened. The argument follows broadly the same pattern as in Case 1, except that
the expression R(n) is useless when q = 5 and ineective when q = 7.
build [see also: construct, establish, develop, make] This enables discontinuous derivations to
be built. The assumption that Q is not a torsion point is built into (5). Kirk, building on
work of Penot, developed a more abstract version of..... Thus A can be written as a sum of
functions built up from B, C, and D.
but [see also: though, although, however, nevertheless, yet, none, other] If..... then (1) holds
but not (2). All but a nite number of the G
s
are empty. This will be proved by showing that
H has but a single orbit on M. However, (ii) is nothing but the statement that..... [= only the
statement that]
by By continuity By the continuity) of f,..... By assumption,..... By the above,..... By the
induction hypothesis,..... By and large, we shall use the same notation as in..... [= In general]
term-by-term dierentiation a 3 by 4 matrix = a 3 4 matrix Thus we need only alter our
constants by a factor of 2 to deal with this case. By quasi-equation we understand a sentence
of the form.....
bypass In order to bypass some technicalities, we also assume that.....
by-product [see also: incidentally, pass] As a by-product, we obtain an explicit formula for.....
C
calculate [see also: compute] Calculating V is an application of Theorem 5. It can be easily
calculated by hand that the nonzero solutions are..... To calculate (2), it helps to visualize the
S
n
as the successive positions in a random walk.
calculation [see also: computation] The calculation of M(f) is usually no harder than the
calculation of N(f). A short calculation shows that..... These upper bounds are too large to
be useful in computer calculations in general, but the ideas in the proofs will surely contribute
to better bounds in the future. We can do a heuristic calculation to see what the generator of
x
t
must be. These calculations can be performed entirely in terms of the generators of G.
By detailed numerical calculation one may show that.....
call [see also: name, designate, term, appeal, invoke, require] A set D is called a phase diagram
if..... [No comma before if here.] The sequence a
n
is what is sometimes called a recovery
sequence for v. Call a set a phase diagram if..... Call H the width of g. [Note that this phrase
denes the last term appearing (width), and H is assumed to be known; cf. Denote by H the
width of g, which denes the notation H.] The lemma motivates our calling R a generalized
22 JERZY TRZECIAK
Picard group. This corollary calls for some explanation and comment. We will call on this
version of the inverse theorem when we come to our applications in Section 2. Adding E to
both sides of (1), we can call upon (2) to obtain (3).
can [see also: able, could, may] No point of T can be a regular point of f. There cannot be
two edges between one pair of vertices. It cannot be that there exists x F, for otherwise
(
x
) =
E
(
x
) = 0, which is not the case. But (1) can be interpreted to mean that.....
cancel [see also: compensate, oset] Because of this, W is never long enough to cancel with M
in the product ABC. We can immediately use one of these poles to cancel out the eect of the
pole at s = 0.
capture The signicance of this fact for our purposes is captured by Corollary 3. The notion
of turbulence almost exactly captures the combinatorial content of not being an S-action.
cardinality a set of cardinality The set E has the cardinality of the continuum.
care [see also: caution] This procedure can be extended to take care of any number of terms.
One should take great care with..... However, (9) needs handling with greater care. The only
point that requires extra care is the verication of..... Special care needs to be exercised in
forming the Cauchy kernel of G(x). The above denition was rst given in [D]; care is required
because this term has been used in a slightly dierent sense elsewhere.
careful The author thanks H. Miller for a careful reading of an earlier draft.
carefully By carefully examining the relations between the quantities U
i
, we see that.....
carry [see also: move, make, perform] This map carries lines to lines carries M onto N).
Each set A carries a product measure. The solutions can be carried back to H(V ) with the
aid of the mapping function . The detailed analysis of..... is carried out in Section 2. In
order to carry out the construction, we must make a judicious choice of P. The method of
proof carries over to domains satisfying..... The method sketched in Section 3 of [Con] carries
through with our choice of =
1
+
2
, but there is one extra ingredient worthy of mention.
case [see also: event, possibility] We rst do the case n = 1. This argument also settles the
case of K = . We nish by mentioning that, suitably modied, the results of Section 2 apply
to the AP case. Note that (4) covers the other cases. There are several cases to consider:.....
We close this article by addressing, in part, the case of what happens if we replace the map
T by convolution. There are quite a number of cases, but they can be described reasonably
systematically. The general case follows by changing x to xa. This abstract theory is not in
any way more dicult than the special case of the real line. Important cases are where S =.....
This case arises when..... Both cases can occur. To deal with the zero characteristic case,
let..... Then either....., or..... In the latter former) case,..... In the case of nite additivity, we
have..... In the case of n 1 In case n 1),..... [Better: If n 1 then.....] In the case where
A is commutative, as it will be in most of this paper, we have..... We shall assume that this
is the case. It cannot be that there exists x F, for otherwise (
x
) =
E
(
x
) = 0, which
is not the case. Unfortunately, this is rarely the case. However, it need not be the case that
V > W, as we shall see in the following example. Such was the case in (8). The L
2
theory has
more symmetry than is the case in L
1
. Note that some of the a
n
may be repeated, in which
case B has multiple zeros at those points. This is handled by a direct case-by-case argument.
MATHEMATICAL ENGLISH USAGE 23
a classic textbook/typical/borderline/simple/extreme/rare/striking) case We use upper case
letters to represent inverses of generators. [= capital letters]
category a set whose complement is rst category a rst category set Dene the category
of..... to have as objects pairs of....., and as morphisms..... The assignment of K
1
to K, and
of T
1
to T, denes a functor between the category of commutative algebras and the category of
compact semigroups with continuous homomorphisms. The most frequently used models fall
into one of the following two categories.
cause [see also: result] The hypothesis f(0) ,= 0 causes no harm in applications, for if.....
Clearly, there cannot be more than one such F, since that would cause E to be non-Hausdor.
caution [see also: beware, care] 1 For measures that are not of nite variation, more caution is
required. Nevertheless, in interpreting this conclusion, caution must be exercised because the
number of potential exceptions is huge.
2 The reader is cautioned that our notation is in conict with that of [3].
centre [amer. center] a ball with centre at x [Or: with centre x, centred at x] In (3.6), the
most favorable case is when t is in the centre of the interval.
certain The spectrum of T was dened in [4] and identied with the spectrum of a certain
algebra A
T
. It seems plausible that..... but we have been able to establish this only in certain
cases. under certain conditions
certainly [see also: surely] This is certainly reasonable for Algorithm 3, given its simple loop
structure. Hence both decay exponentially as x , therefore certainly remain bounded.
As Corollary 2 shows, it is certainly a question deserving further exploration. Here, of course,
the set A produced is rather thin and certainly nowhere near the densities we are looking for.
challenge We shall be brief here and just estimate the terms that are the most challenging.
chance They have a 90 per cent chance of success.
change 1 [see also: adaptation, adjustment, modication] The shape of F undergoes radical
changes as x moves from A to B. The change in arg z around is..... A change in per-
spective allows us to gain not only more general, but also ner results than in [ST]. After
the change of variable z =..... Theorems 3 and 6 of [2], with the appropriate changes, are
also valid. The rest of the proof goes through as for Corollary 2, with hardly any changes.
The present proof is so arranged that it applies without change to holomorphic functions
of several variables. From (ii), with an obvious change in notation, we get..... Properties
involving topological centres are unaected by a change to an equivalent weight. a slight
small/subtle/considerable/radical/profound/major/substantial) change
2 [see also: adapt, adjust, alter, convert, modify, replace, revise, transform, vary] Reversing
the orientation of Y changes B into B. By changing at least one of the circles to a rectangle,
we obtain..... The method is to change variables to y = x + c so that..... The parameter
of the model is changed from p to 1 p. If..... then A changes from (b, c) to (c, d). The
dierence can change only by an even integer. When we pass from T to T + , f(z) changes
by approximately M. The limit in 4(b) is unchanged if g is replaced by f. These denitions
are taken almost unchanged from the papers of Furstenberg [2, 3].
character Properties (a) to (c) are analytic in character.
24 JERZY TRZECIAK
characteristic 1 [see also: feature, aspect, detail, ingredient, point] The two characteristics are
connected, but the relationship is quite a complex one.
2 [of sth; see also: typical, peculiar] This property is characteristic of functions in Q.
characterization Neumanns characterization of the elements which induce an indecomposable
operator is as follows:..... In our next theorem, we state a characterization of..... which does
not seem to have been noticed previously.
characterize This condition completely characterizes A
p
when..... The chapter concludes with
a theorem which shows that the maximum property almost characterizes the class of holomor-
phic functions. Most distribution spaces can be characterized through the action of appropriate
convolution operators. Our main results state in short that MEP characterizes type 2 spaces
among reexive Banach spaces. More generally this argument also applies to characterizing
Hurewicz subsets of I.
check 1 [see also: examination, inspection, verication] It is an elementary check that A is a
vector space.
2 [see also: examine, verify, test] It is easily checked that f is l.s.c. The continuity of f is
readily checked, and we can then apply Moreras theorem. The only thing to check is that
maps compose correctly; but this is an easy, if tedious, verication.
choice [see also: selection] Thus F is independent of the choice of the family S. Indeed, it
is routine to verify that the index so constructed is independent of the choices made. We do
not know how V depends on the various choices made. The choice of A is clearly irrelevant, so
assume A = 0. Our choice of V shows that..... With this choice of b,..... For this choice of
, and with u = z = s, the expression (5.3) simplies greatly. Corresponding to each choice
of V there is a function f such that..... The problem is that, whatever the choice of F, there
is always another function f such that..... This just amounts to a choice of units. By choice
of V ,..... Having established (1), one might be tempted to try to extend this result to general
p through the choice of a suitable ideal B. There are O(1) possible choices for x. Where
there is a choice of several acceptable forms, that form is selected which..... It is the freedom
of choice of D in this construction that enables us to..... On the other hand, there is enormous
ambiguity about the choice of M. In order to carry out the construction, we must make a
judicious choice of P. By revising our choice of A if necessary, we may assume that.....
choose [see also: select, pick] This implies that however we choose the points y
i
, the intersection
point will be their limit point. In fact, we are at liberty to choose the function f, so we set
f =..... This also resolves the ambiguity introduced earlier in choosing an order of the lifts
of U. Where we could, we have chosen these examples from naturally occurring mathematical
structures. [Not: choosen; also, note the double r in occurring.] .....where C is so chosen
chosen so) that..... .....where C can be chosen independent of n. Here a and b are chosen to
satisfy..... Here Y is a Poisson variable suitably chosen to approximate X in distribution. If
A is such an operator (appropriately chosen) then..... However, the space Z depends crucially
on the particular ambient algebra A chosen.
circle What is F(c) if c is a positively oriented circle? a circle with centre a and radius r
circumstances [see also: state, situation, condition] Only in exceptional circumstances is it
true that f(x +y) = f(x) +f(y). [Note the inversion.] The idea is that C is xed, but X and
Y vary according to circumstances. Under In) what circumstances can this happen?
MATHEMATICAL ENGLISH USAGE 25
circumvent [see also: get around, overcome] The assumptions imposed on f circumvent mea-
surability diculties.
cite [= mention; see also: quote] In the proof of the cited theorem, we found sequences a and b
with certain properties. the paper cited on p. 45 the above-cited paper
claim 1 To establish the last claim, let..... Their remarkable achievement seemed to validate
Johns claim. However, it soon turned out that.....
2 [see also: assert] Lax claimed to obtain a formula for....., but the argument turned out to be
incomplete. Note that for semilattices one could not claim that A is algebraically closed in B.
We claim that, by setting w to zero on this interval, the value of F(w) is reduced. So F must
be constant on H, thereby showing that A = B, as claimed desired/required).
clarity When clarity requires it, we shall write..... We shall write a.e. [] whenever clarity
requires that the measure be indicated. [Note the subjunctive be.] For the sake of clarity,
we shall indicate in what follows to which space X belongs. It seems preferable, for claritys
sake, not to present the construction at the outset in the greatest generality possible. Standard
Banach space notation is used throughout. For clarity, however, we record the notation that is
used most heavily.
class Let F be of class T
k
for some k. Then E is an equivalence relation on S. Each E-class of
S contains..... We divide them into three classes depending on the width of.....
classic [see also: classical] Chapter 2 of the classic text [6] by R. Nevanlinna has a detailed
treatment of this construction. [Note the dierence between classic and classical : classic
means of acknowledged excellence or remarkably typical (a classic example); and classical
means well-established, not modern.]
classical [see also: classic] Theorems 1 to 3 are classical results, due to Lebesgue. Since most
of the results presented are quite classical (the novelty lies in the arrangement, and some of the
proofs are new), I have not attempted to document the source of every item. The following
theorem is clearly motivated by the classical LP-decomposition. Our proofs make substantial
use of classical topology of the plane. [Note the dierence between classic and classical : classic
means of acknowledged excellence or remarkably typical (a classic example); and classical
means well-established, not modern.]
classication The middle part of Table 2 compares the classication according to max a
i
, where
only the longitudinal information is utilized, with those according to max b
ik
, where both lon-
gitudinal and survival information are used.
clear [see also: evident, obvious, plain] The denition of M makes it clear that F is continuous.
[Note that the it is necessary here.] Now (d) is clear from the very last statement of Theorem 4.
The diculty is that it is by no means clear what one should mean by a normal family. It
is also clear that there are extensions to....., but they do not seem to be worth the eort of
formulating them separately. It is intuitively clear that the amount by which S
n
exceeds zero
should follow the exponential distribution. If G is clear from context, then we suppress reference
to it in the notation. However, it is unclear how to prove Corollary 3 without the rank theorem.
clearly [see also: evidently, manifestly, obviously] It would clearly have been sucient to as-
sume..... It seems that the relations between these concepts emerge most clearly when the
26 JERZY TRZECIAK
setting is quite abstract, and this (rather than a desire for mere generality) motivates our ap-
proach to the subject.
close 1 [see also: nish, conclude] We close this section with a discussion of..... We close this
article by addressing, in part, the case of what happens if we replace the map T by convolution.
In closing this section we take up a result which will play a pivotal role in the characterization
of..... We close this o by characterizing.....
2 [see also: near] We have to check that F does not get too close to p. for t close to 0 Part
of the conclusion is that F moves each z closer to the origin than it was. Let P be a point of
U closest to Q. It is this point of view which is close to that used in C

-algebras. See [AB]

for a proof that is close to the original one. The above bound on a
n
is close to best possible
to the best possible). The products F
i
G
i
are very close to satisfying (1). How close does
Theorem 1 come to this conjecture? These results therefore describe the very close connection
between the method of encoding and the structures we are aiming to classify.
closely [see also: intimately] Our results are closely related to those of Strang [5]. This notion
is closely connected with that of packing dimension. The proof follows very closely the proof
of (2), except for the appearance of the factor x
2
. There are other problems with this example
which would hinder any attempt to follow the proof given here too closely. The proof closely
parallels that of Theorem 1. The spectral theory of..... is closely linked to the discrete Hilbert
transform.
clue A complete explication of the Fox spaces is warranted, as it will likely reveal further clues
to the dierences between the parabolic and hyperbolic theories.
cluster The observed values of X will on average cluster around points where.....
coecient Let D be the coecient of x
n
in the expansion of f. [Not: coecient by x
n
]
Equating coecients we see that..... Equating the coecient of x
2
in V to zero, we get.....
Now equate the coecients of x
2
at either end of this chain of equalities.
coincidence This is no coincidence, in the light of the remarks preceding Denition 2. However,
this equality turned out to be a mere coincidence.
collect [see also: combine, gather] Collecting the terms with even powers of x, we obtain.....
collection [see also: set] Instead of dealing with lines one by one, we deal with collections of
lines simultaneously. Neighbourhoods of points in these spaces appear at rst glance to have
a nice regular structure, but upon closer scrutiny, one sees that many neighbourhoods contain
collections of arcs hopelessly folded up. Continuing in this fashion, we get a collection V
r
of
open sets, one for every rational r, with..... Next we relabel the collection A
n
, B
n
as C
n
.
The words collection, family and class will be used synonymously with set. The theorem
implies that some nite subcollection of the f
i
can be removed without altering the span.
column We tabulate the outcome for n 10; in particular, the column headed R lists R(n)
truncated to three decimal places.
combine [see also: collect, couple, conjunction, gather, piece together] If we combine this with
Theorem 2, we see that..... Theorems 1 and 2 combined give a theorem of Fix on..... Now
(1) is just (2) and (3) combined. Theorems 1 and 2 combine to give the following result.
Then (5) combined with (6) gives..... On combining this with our other estimates in (3.5), we
MATHEMATICAL ENGLISH USAGE 27
deduce that..... As noted in the introduction, this is basically combining Sawyers result with
a variation of the arguments of Hunt.
come ....., the last inequality coming from (5). This argument comes from [4]. If we restrict
A to sections coming from G, we obtain..... Our interest in..... comes from the fact that.....
We now come to a theorem which was rst proved by..... We will call on this version of
the inverse theorem when we come to our applications in Section 2. Two necessary conditions
come to mind immediately. How close does Theorem 1 come to this conjecture? The algebra
D(F) comes equipped with a dierential d such that H(d) = 0. Along the way, we come across
some perhaps unexpected rigidity properties of familiar spaces. Thus, everything comes down
to proving the existence of M. This is where the notion of an upper gradient comes in. Other
situations in dynamics where the p-adic numbers come up are surveyed in [W].
commence [see also: start, begin] The paper was commenced whilst the second author held a
Fullbright Fellowship.
comment [on sth; see also: remark] He makes the following comments:..... This corollary calls
for some explanation and comment. The author thanks John Fox for helpful comments on the
content and presentation of this paper. The author gratefully acknowledges the referees helpful
comments pertaining to the rst draft of this paper. We shall encounter similar situations again,
and shall apply convergence theorems to them without further comment.
common [see also: customary, familiar, usual] The functions f
i
(i = 1, . . . , n) have no common
zero in . Then F and G have a factor in common. It has some basic properties in common
with another most important class of functions, namely, the continuous ones. Take g
1
, . . . , g
n
without common zero. Denote by the angle at x that is common to these triangles. Here
we use an inductive procedure very common in geometric model theory.
commonly [see also: usually] If f maps D into itself, then T is commonly called a composition
operator.
companion In a companion paper [5] we treat the case..... here is a companion result to
Theorem 2 which shows that.....
comparable Thus will be less than by an amount comparable to a(s).
compare [with sth; see also: relative] The algorithm compares x with each entry in turn until
a match is found or the list is exhausted. Our asymptotic results compare reasonably well
with the numerical results reported in [8]. We divide them into three classes depending on
the width of the middle term as compared with the width of the other terms. We show that
A is negligible compared with B. Quarterly sales rose by 9% compared with the same period
in 2007. A computer virus can be compared to a biological virus. [Use compare with when
you are contrasting two things, and compare to when you want to liken them.]
comparison Comparison of (2) and (3) gives.....
compensate [see also: cancel, oset] We can immediately use one of these poles to compensate
for the eect of the pole at s = 0.
complement We denote the complement of A by A
c
whenever it is clear from the context with
respect to which larger set the complement is taken. [Not: compliment]
28 JERZY TRZECIAK
complete 1 [see also: whole, entire, full, total] The proof of the lemma is complete. [Not:
completed; cf. next example]
2 [see also: accomplish, nish] The proof is completed by invoking Theorem 5. This completes
our argument for (1). ....., which completes the proof. [Not: what completes the proof]
completely [see also: entirely, fully, wholly] We remark that Bennett and the third author
completely solved equation (4) in the case when..... The intervals we are concerned with are
either completely inside A or completely inside B.
completeness We include the proof for completeness. For completeness of exposition, we now
recall the denition of the ordering on R.
completion Mary Lane deserves our special thanks for her part in bringing this volume to a
successful completion.
complex [see also: complicated, dicult, hard, involved, intricate, sophisticated] The two char-
acteristics are connected, but the relationship is quite a complex one. In detail the classication
is complex, but in essence it is simple.
complicated [see also: complex, dicult, elaborate, hard, involved, intricate] However, for
three or more subgroups, matters become more complicated. The known results show that
the situation is innitely more complicated there than in L
2
. Although the idea is simple, its
implementation is complicated by the fact that..... We omit the details, which are notationally
complicated.
complication [see also: diculty, problem, trouble] However, as we are about to see, this
complication is easily handled. A further complication arises from BP, which works rather
dierently from the other labels.
compose [see also: contain, comprise, consist, constitute, make up] Illiterates compose 20% of
the population. Each committee is composed of ten senators. [Not: is comprised of]
composition Then G is a group with composition as group operation. composition on the right
with p
comprehensive For a comprehensive treatment and for references to the extensive literature on
the subject one may refer to the book [M] by Markov.
comprise [see also: compose, consist, contain, constitute, make up, include] The monograph
comprises three parts. [= has three parts only; cf.: The monograph includes two chapters on
divergent series. Use comprise only when you mention all parts.] [The form is comprised of
(instead of is composed of) is considered incorrect by many native speakers.]
computable This solution has the additional advantage of being easily computable.
computation [see also: calculation] Incidentally, our computation shows that..... A short
computation shows that..... It is sucient to make the computation for T. In fact, the
computation for A becomes somewhat simpler. One of the appealing aspects of the spectral
set is that it readily lends itself to explicit computation. This proves one half of (2); the other
half is a matter of direct computation. However, one needs to perform computations which
become time consuming as n increases.
MATHEMATICAL ENGLISH USAGE 29
computational A computational restraint is the algebraic number theory involved in nding
these ranks, which will typically be more demanding than in our example of Section 1.
compute [see also: calculate] By computing the second derivative we note that x = 1 is a
maximum point for f. Find integral formulas by means of which the coecients c
n
can be
computed from f. This provides an eective means for computing the index. Computing
f(y) can be done by enumerating A(y) and testing each element for membership in C.
conceivably [see also: possibly] Conceivably, S may also contain other sets.
concentrate In this section, we concentrate on those maps that satisfy..... [Not: We are
concentrated on those maps] the unit mass concentrated at x
concept [see also: idea, notion] We now turn to a brief discussion of another concept which is
relevant to Johns theorem. It seems that the relations between these concepts emerge most
clearly when the setting is quite abstract, and this (rather than a desire for mere generality)
motivates our approach to the subject. They dened the concept solely in terms of the norm
of the Banach space, deliberately avoiding any extra structure.
conceptually We remark that this proof seems conceptually and calculationally simpler.
concern [see also: regard, relate, pertain, involve] 1 Our next concern will be the behaviour of
P(r) as r tends to 0.
2 It is the connection between..... that will concern us here. This book is wholly concerned
with..... We now introduce the algebras we shall be concerned with. The question we are
concerned with is whether or not f is C
1
. each of the spaces concerned In what follows,
we shall concern ourselves only with case (a). For relevant background material concerning
random walks, see [2]. Functions which are equal almost everywhere are indistinguishable as
far as integration is concerned. Concerning (i) Regarding (i)/For (i)), we rst prove that.....
concise [= succinct; see also: brief] in a concise manner
concisely More concisely, we consider the measurable space (X, B).
conclude [see also: close, end, nish, complete, deduce, infer] We conclude this section with a
useful lemma. We conclude with two simple lemmas to be used mainly in..... This concludes
the verication of Claim 2. The proof concludes by observing that..... The chapter concludes
with a theorem which shows that the maximum property almost characterizes the class of
holomorphic functions. We can now integrate n times to conclude that (1) indeed holds. We
conclude from Theorem 4 that A is Arens regular.
conclusion [see also: assertion, statement] The rst conclusion is immediate from Theorem 1.
This conclusion does not follow if the condition f A is omitted. Part of the conclusion
is that F moves each z closer to the origin than it was. One might hope that in the particular
case of the GL energy, this could be established, but we do not see an easy path to such a
conclusion. ....., which is the desired conclusion. Actually, the proof gives an even more
precise conclusion:..... Identical conclusions hold in respect of the condition BN. [= concerning
BN] Here are some other situations in which we can draw conclusions only almost everywhere.
However, a few tentative [= cautious, not rm] conclusions can be drawn. We shall see that
the two cases where A = 1 and where B = 2 give contrasting conclusions.
concrete [see also: precise, specic] We can now restate Theorem 1 in concrete terms:.....
30 JERZY TRZECIAK
condition [see also: assumption, requirement] We prove, under mild conditions on f, that.....
In this section we investigate under what conditions the converse holds. They established
the Hasse principle subject to a rank condition on the coecients. Take N to be a family of
normal measures in P(X) such that N is maximal subject to the condition that the supports
of the measures in the family are pairwise disjoint. This is a condition on how large f is.
The next theorem provides conditions for the existence of..... The corollary gives a necessary
and sucient condition on p for g to belong to A
p
. A necessary and sucient condition for
A to be open is that C be closed. Then (3.5) is a necessary and sucient condition for there
to be a function f such that..... This condition also turns out to be necessary. It should
be no surprise that a condition like a
i
,= b
i
turns up [= appears, shows up] in this theorem.
Condition (c) is intended to give us rm control over..... Finally, we must check that our F
satises condition (2) of Theorem 1. [Not: veries condition (2)] Let A be the subset of
X where this coboundary condition obtains. [= is satised] We can also appeal to Lemma
5 to see that the uniform continuity condition (5.3) is met. When the rank condition fails to
hold we use (3) instead. We note that H is in fact not Lipschitz continuous if this condition is
violated. a restrictive stringent) condition
conditional Dene a
k
to be the probability that exactly k out of the 2n values X
i
exceed T,
conditional on X
0
> T. Here Q
j
for j = 1, ..., n are drawn independently, conditional on the
values generated at level m.
conduct [see also: carry out] This work was conducted during a visit of the author to Dartmouth
College.
conference In 2000, two important number theory conferences were held at Princeton University.
conne [see also: limit, restrict] We conne ourselves to discussing the case..... For simplicity,
we conne attention to radial moments.
conrm This interpretation is conrmed by (4). Theorem 4 conrms conjecture H when B = 1.
conict 1 The reader is cautioned that our notation is in conict with that of [3].
2 Unfortunately, the notation from number theory slightly conicts with the notation from
probability theory.
confound [see also: confuse] This group will be designated by E
3
(not to be confounded with
Euclidean 3-space).
confuse [see also: confound] Encode the h-vector as a polynomial h(x, y) in noncommuting
variables x and y (not to be confused with the indices in the previous section).
confusion It will cause no confusion if we denote..... If no confusion can arise, we write K for
both the operator and its kernel. We drop the subscript when confusion is unlikely.
conjecture 1 [see also: hypothesis] The conjecture will be disproved by exhibiting..... In
addition to illustrating how our formulas work in practice, it provides a counterexample to
Browns conjecture. The conjecture (now known not to be true in general) was that..... a
long-standing conjecture The as yet unproved conjecture of Newman is that..... Conjecture 2
of [KH], to the eect that [= meaning that] there is no relation P with E(P) = 1, still remains
open. Some partial evidence to support this conjecture is discussed in [3]. We oer numerical
evidence to support a conjecture that there exist innitely many primes of this type. All the
MATHEMATICAL ENGLISH USAGE 31
evidence points to the validity of the conjecture. How close does Theorem 1 come to this
conjecture?
2 It is conjectured in [B3] that.....
conjunction [see also: combine, together] This, in conjunction with (6), yields..... For later
use in conjunction with the weighted averages occurring in (2), we next consider..... [Note the
double r in occurring.]
connect [see also: join, link, relate, tie, associate] the segment connecting joining) a to b This
notion is closely connected with that of packing dimension. [Use connect to when you mean
join, and connect with when you mean relate.] The two characteristics are connected, but
the relationship is quite a complex one.
connection [see also: link, relationship] It is the connection between..... that will concern us
here. These results therefore describe the very close connection between the method of encoding
and the structures we are aiming to classify. The problem has a very natural connection with
the problem of the distribution of the zeros of a bounded holomorphic function in a half-plane.
However, the connection with Gromovs work has been obscured in recent years by an emphasis
(in the algebraic topology literature) on conguration spaces. Note in connection with (iv)
that..... In this connection, we remark that.....
consecutive [= successive] The elements of each array must occupy consecutive memory loca-
tions. Suppose the rst three characters of the pattern match three consecutive text characters.
Its role is to rule out having two or more consecutive P-moves (on the grounds that they can
be performed in one go).
consequence [see also: result, eect, outcome, implication] This is a consequence of Dinis
theorem. It is a consequence of the HahnBanach theorem that..... An obvious consequence
of Theorem 2 is the following. As a consequence of Lemma 2, there is..... This has numerous
important consequences; to name just two, Brauer proved that..... This is impossible in conse-
quence of the last corollary. This article features results in both spectral theory and operator
ergodic theory made possible by a recent renewal of interest in the consequences of Jamess
inequalities.
consequently [see also: hence, thus, therefore, so] Consequently, A has all geodesics closed if
and only if B does.
consider [see also: account, discuss, address, deal, regard, think, take up, treat, view] We shall
be considering L on various function spaces. All our results can be extended in this way, but
we shall stick to considering P rather than P

. Here U is considered viewed/regarded) as

acting on M. An arc is considered a degenerate star. There are several cases to consider:.....
For size 1 this makes no dierence, but for sizes 2 and 3 it considerably cuts down the number
of cases to be considered. We consider every subset of N, whether nite or innite, to be
an increasing sequence. The product being considered grows like n
3
. Here the constants of
proportionality depend on the particular curve being considered. We begin by describing the
class of functions f considered, which includes the special cases quoted above.
considerable [see also: important, signicant, substantial] However, they now dier by a con-
siderable amount. It is possible that the methods of this paper could be used to....., but there
remain considerable obstacles to overcome.
32 JERZY TRZECIAK
considerably [see also: greatly, signicantly, substantially] The performance of the device has
improved considerably. Furthermore, adopting this strategy considerably eases constructing a
coding tree from a linear order. For size 1 this makes no dierence, but for sizes 2 and 3 it
considerably cuts down the number of cases to be considered. Theorem 3 is remarkable in that
considerably fewer conditions than in the previous theorems ensure universality.
consideration [see also: check, discussion, examination, scrutiny, treatment, verication] In
this section we return to our general consideration of p(R). It now follows from the maximum
modulus theorem that F
n
(A) A, and consideration of F
n
shows that actually F
n
(A) = A.
We make two standing assumptions on the maps under consideration. A moments consid-
eration will show that..... For these considerations it will be convenient to state beforehand,
for easy reference, the following variant of..... The same considerations show that..... By
symmetry considerations, it is sucient to search over a region in which.....
consist [see also: comprise, compose, make up, contain, lie] This group consists of the elements
of GL(n) of determinant 1. [= is made up of] Let B() be the quotient of B
b
() by the
closed linear subspace consisting of the functions which are zero outside a meagre subspace of
. The subspaces of M() consisting of the discrete and continuous measures are M
d
() and
M
c
(), respectively. We dene the set T to consist of those f for which..... Let R consist of
all z = x +iy such that [x[ < 1. But M does not consist of 0 alone. The inverse image of A
consists of just the basepoint of X. The proof consists in the construction of..... [= The main
feature of the proof is] The diculty consists in generalizing (b).
constant Here c denotes a constant which can vary from line to line. The least such constant
is called the norm of G. Unfortunately, F is dened only up to an additive constant. The
implied constants The constants implicit in the symbol ) depend only on r.
constantly the constantly zero function
constitute [see also: form, make up, represent, account for] Nitrogen constitutes 78% of the
earths atmosphere. Volunteers constitute 90% of the Units work force. 52 cards constitute
a pack.
constraint [see also: limitation, restraint, restriction] The idea is to relax the constraint of
being a weight function in Theorem 3.
construct [see also: build, establish, develop, create, make] Then F is constructed from the F
j
.
Although [1] deals mainly with the unit disc, most proofs are so constructed that they apply
to more general situations. Indeed, it is routine to verify that the index so constructed is
independent of the choices made. Thanks to Lemma 2, we can now modify the proof of
Theorem 3 by inductively constructing ve sequences.....
construction Actual construction of..... may be accomplished in a variety of ways. The fol-
lowing construction can be carried out. Similar constructions may be made for other n odd.
We can make a similar construction to form E
n
.
consult We encourage the reader unfamiliar with techniques from the theory of..... to con-
sult [BS].
contain [see also: comprise, include] That (2) implies (1) is contained in the proof of Theorem 1
in [4]. The preceding proof contains a result which is interesting enough to be stated separately.
MATHEMATICAL ENGLISH USAGE 33
Section 2 contains an overview of the necessary background.
content That (2) implies (1) is the content of Corollary 3. The notion of turbulence almost
exactly captures the combinatorial content of not being an S-action. The author thanks John
Fox for helpful comments on the content and presentation of this paper. Intuitively, entropy
of a partition is a measure of its information contentthe larger the entropy, the larger the
information content.
context [see also: framework, setting, set-up] in the present context In the physical context
already referred to, K is the density of..... [Note the double r in referred.] Berg spaces have
been rarely considered outside the metric context. The context will make it clear whether S
denotes a permutation of M or the corresponding symmetry. When the target space is clear
from context, we just write..... If G is clear from context, then we suppress reference to it in the
notation. Strictly speaking, we should write something like a(l, m, n) to reect the dependence;
we shall rely upon context instead. To place Theorem 1 in context, consider two real vector
elds.....
continue [see also: extend, go on, proceed, pursue, remain, persist, stick, still] Suppose that the
process continues indenitely. Continuing in this fashion, we get a collection V
r
of open sets,
one for every rational r, with..... Continuing in this manner, we see that for 4, the curve
C
4k+3
is found in the half-strip B. This recursion continues, producing a sequence of nite
sets such that..... By abuse of notation, we continue to write f for f
1
. Since Z is a nite set,
we may continue subtracting suitable scalar multiples of the x
i
from x. Let us continue with
the proof of Theorem 2. An obvious question to ask is whether the assertion of Theorem 1
continues to hold for..... This paper, for the most part, continues this line of investigation.
This work continues research begun in [5].
continuous We note that H is in fact not Lipschitz continuous if this condition is violated.
a function continuous in space variables More precisely, f is just separately continuous. The
map f, which we know to be bounded, is also right-continuous. a function continuous from
the right We follow Kato [3] in assuming f to be upper semicontinuous. Examples abound
in which P is discontinuous.
continuum The set E has the cardinality of the continuum. Suppose X is a separable space
of cardinality continuum. However, F has continuum many components.
contradict [sth; see also: contrary] This contradicts the fact that..... [Not: contradicts to the
fact] ....., which contradicts F being countable. But this is contradicted by Proposition 3
applied to g.
contradiction [see also: contrary, otherwise] To obtain a contradiction, we suppose that.....
Suppose, to derive a contradiction, that..... Striving for a contradiction, suppose that..... Aim-
ing for a contradiction, suppose that..... Suppose, towards a contradiction for contradiction),
that..... ....., which is a contradiction. Now we have the required contradiction since.....
This leads to the contradiction that 0 < a < b = 0. ....., in contradiction with Lemma 2.
This is a clear contradiction of the fact that.....
contrary [to sth; see also: otherwise, contradict, contradiction] Suppose that, contrary to our
claim,..... Suppose the contrary and choose..... We now show that A is closed. Suppose that,
on the contrary, there is an x..... In the contrary case,..... While nonparametric priors are
34 JERZY TRZECIAK
typically dicult to manipulate, we believe the contrary is true for quantile pyramids. Hence
F is bounded, contrary to assumption. There is no evidence to the contrary. [Do not write:
Contrary to [5], our example is nite-dimensional if you mean: In contrast to [5], our example
is nite-dimensional.]
contrast 1 [to/with sth; see also: dierence, oppose] Note that, in contrast to F, the function
G is bounded. [Not: contrary to F] In contrast, Theorem 2 shows that..... By contrast, T
does not have this symmetry. This is in marked contrast to the behaviour of orthonormal sets
in a Hilbert space. Note the contrast with Theorem 3.
2 [with sth; see also: dier] This contrasts sharply with the situation in metrizable spaces.
This contrasts with (but does not contradict) Theorem 2 of [6]. We shall see that the two cases
where A = 1 and where B = 2 give contrasting conclusions.
contribute [sth; to sth] The terms involving a ,= 1 contribute O(n), since..... These upper
bounds are too large to be useful in computer calculations in general, but the ideas in the proofs
will surely contribute to better bounds in the future.
contribution [from sth; to sth] This decrease is oset by the contribution from the poles. Thus
we may count the contribution to the periodic points from each prime separately. We shall
thus avoid doubly counting the same contribution. Note that A
n
makes only a contribution of
at most N
2
. Clearly, the contribution from those r with A(r) > 0 can be neglected. However,
we prefer to avoid this issue altogether by neglecting the contribution of B to S. If nothing
else, I hope to convince my readers that Segals theorem deserves recognition as a profound
contribution to Gaussian analysis.
control 1 This gives some control over the behaviour of..... Condition (c) is intended to give us
rm control over..... This extension retains control on..... at the sacrice of losing some control
on..... [Not: loosing] The requirement that competitors are symmetric will only be used to
get some very rudimentary control on the asymptotics of G.
2 The obstacle we have is that we must control the area of G.
convenience For convenience we ignore the dependence of f on g. For convenience of exposition,
we work with an error term of the form..... We note that the assumption of GCH is made for
convenience and ease of presentation. For the convenience of the reader, we repeat the main
points. [Or: For the readers convenience; not: For the commodity of the reader]
convenient [see also: appropriate, suitable, useful, suit] This realization is particularly con-
venient for determining..... It is convenient to view G as a nilpotent group. In this case it
is natural (and notationally convenient) to pick W to be..... For these considerations it will
be convenient to state beforehand, for easy reference, the following variant of..... Therefore,
whenever it is convenient, we may assume that..... We identify A and B whenever convenient.
We shall, by convenient abuse of notation, generally denote it by x
t
whatever probability space
it is dened on. We shall nd it convenient not to distinguish between two such sequences
which dier only by a string of zeros at the end.
conveniently These are conveniently divided into three disjoint sets. The proof conveniently
splits into two cases.
convention This convention simplies the appearance of results such as the inversion formula.
We adhere to the convention that 0/0 = 0. We adopt throughout the convention that compact
MATHEMATICAL ENGLISH USAGE 35
spaces are Hausdor. We adopt the convention that the rst coordinate i increases as one goes
downwards, and the second coordinate j increases as one goes from left to right. We make
the convention that f(Q) = i(Q). We maintain the convention that implied constants depend
only on n. We regard (1) as a mapping of S
2
into S
2
, with the obvious conventions concerning
the point . By convention, we set a(x, y) = 0 if no such spaces exist. This sort of tacit
convention is used throughout Gelfand theory.
converge [see also: tend, approach] Then F
n
converges simply uniformly/weakly/weak

/in the
norm of L
p
/in L
p
norm/in norm/in probability).
convergence Moreover, one has estimates on the rate at which this convergence is taking place.
Addressing this issue requires using the convergence properties of Fourier series. The con-
vergence of the sum on the left is of course a weaker statement than the convergence of (2).
We give X the topology of uniform convergence on compact subsets of A. Almost everywhere
convergence is the best we can hope for.
converse [of sth; to sth; see also: inverse, opposite, reverse] In this section we investigate under
what conditions the converse holds. The converse is far from obvious. For the converse,
consider..... This theorem is a converse partial converse) of Theorem 2.
conversely We have thus proved that (a) implies (b). Conversely, assume that (b) holds and
dene..... [Not: Reciprocally, assume.]
convert [see also: transform, change into, turn into, make into] This isomorphism converts the
shift operator to the multiplication operator. The Fourier transform converts multiplication
by a character into translation, and vice versa, it converts convolutions to pointwise products.
convey [see also: express, indicate] The main information conveyed by this formula is that A
cannot be.....
convince If nothing else, I hope to convince my readers that Segals theorem deserves recognition
as a profound contribution to Gaussian analysis.
coordinate There are N vertices of the 24-cube which have a 1 in the ith coordinate and a total
of ve 1s. We adopt the convention that the rst coordinate i increases as one goes downwards,
and the second coordinate j increases as one goes from left to right. Now X can be taken as
coordinate variable on M. This explains why we chose 9 rather than, say, 1 for the second
coordinate. Keep only those vertices whose coordinates sum to 4.
core [see also: essence, heart] At the core of our proof of Theorem 1 is a simple counting
argument.
corollary [to sth; of sth] To illustrate, let us state the following corollary. That (2) implies (1)
is the content of Corollary 3. This corollary calls for some explanation and comment. The
rest of the proof goes through as for Corollary 2, with hardly any changes. The signicance of
this fact for our purposes is captured by Corollary 3. We will nish this section by oering a
second application of our machinery (although in truth it is largely a corollary of the above).
correct [see also: true, valid, hold] Whether or not this is correct does not matter; we are trying
to motivate the proof that follows.
correctly The only thing to check is that maps compose correctly; but this is an easy, if tedious,
verication.
36 JERZY TRZECIAK
correspond [to sth; see also: accord, agree, match, respective] To every f there correspond two
functions a and b. Consider the dierences between these integrals and the corresponding ones
with f replaced by g. Let v be an arbitrary control and x
t
the corresponding controlled process.
[Not: the respective process; respective requires a plural noun.] It seems appropriate to
mention in passing the corresponding formula for the cohomology of B
n
. Corresponding to
each choice of V there is a function f such that.....
correspondence Observe that A is thereby put into one-to-one correspondence with B. The
elements of C are in one-to-one correspondence with.....
coset Here A is a coset of H in G. [H is a subgroup of G.]
cost [see also: expense, price] The latter hypothesis can be removed at the cost of an extra
factor on the right hand side of (5).
could Where we could, we have chosen these examples from naturally occurring mathematical
structures. [Note the double r in occurring.] The number of distinct values that could be in a
memory cell is at most s. This shows that f could not have n zeros without being identically
zero. Without losing any generality, we could have restricted our denition of integration to
integrals over all of X. It is possible that the methods of this paper could be used to....., but
there remain considerable obstacles to overcome. One might hope that in the particular case
of the GL energy, this could be established, but we do not see an easy path to such a conclusion.
count 1 A quick count shows that A has 36 points.
2 Those more than half a square count as whole ones. We shall thus avoid doubly counting
the same contribution. At the core of our proof of Theorem 1 is a simple counting argument.
the number of zeros of f counted according to their multiplicities counted with multiplicity)
counterexample In addition to illustrating how our formulas work in practice, it provides a
counterexample to Browns conjecture. For a counterexample, consider S =.....
couple 1 [see also: some, several] A couple of dierences between this denition and the treat-
ment of [KS] should be noted.
2 [see also: combine] This follows immediately from (8) coupled with the fact that.....
course [see also: process] This paper was written in the course of the semester on..... In the
course of proof, we shall encounter..... In the course of writing this paper we learned that P. Fox
has simultaneously obtained results similar to ours in certain respects. The prerequisite for this
book is a good course in advanced calculus. [Not: of advanced calculus] Then, of course, F
will be one-to-one. Of course, before going further, we must check that such a P exists. The
convergence of the sum on the left is of course a weaker statement than the convergence of (2).
cover [see also: encompass, include] Note that (4) covers the other cases. Then P covers M
twofold.
create [see also: produce, generate, develop] All of the action in creating S
i+1
takes place in
the individual cells of type 2 or 3.
criterion [pl. criteria] Part (c) is a frequently used criterion for the measurability of a real-valued
function. The criterion for its existence is Af = 0. The following is an explicit criterion for f
to be hamiltonian. Note that this lemma does not give a simple criterion for deciding whether
MATHEMATICAL ENGLISH USAGE 37
a given topology is indeed of the form T
f
. The following result gives a criterion for when a
point belongs to B.
critical [to sth; see also: basic, crucial, essential, key] In particular, F is compact (a fact which
was critical to our arguments in [6]). However, with the recent advent of simulation based
inference, the need for analytically tractable posteriors is no longer critical.
cross Let M

be the minor obtained by crossing o the last row and column of M.

crucial [to sth; see also: basic, critical, main, principal, key, important] The following lemma,
crucial to Theorem 2, is also implicit in [4]. The main crucial/key) ingredient in the proof
of..... is.....
crucially However, the space Z depends crucially on the particular ambient algebra A chosen.
cumbersome [see also: heavy] We do not use (5) in our proofs, because it makes the notation
more cumbersome.
curiously But, curiously, the equivalence of (A) and (B) may fail.
customary [see also: common, familiar, usual] As is customary, we use l
n
for C
n
with the
norm..... If....., it is customary to write..... rather than..... With the customary abuse of
notation, the same symbol is used for both F and its r-image.
cut [see also: reduce, diminish, decrease, lower] For size 1 this makes no dierence, but for sizes
2 and 3 it considerably cuts down the number of cases to be considered.
D
data [see also: information, material, detail, evidence] The empirical data quite clearly indicate
that we should expect proportions close to 3/4. [Or: indicates] A model for analysing rank
data obtained from several observers is proposed. for some set of input data
date 1 in about 1885; in the year 2000; as early as 1885; in Hilberts 1905 paper; the revised
1993 edition; the seminal paper of Dixmier [D] of 1951.
2 The best known result to date has been obtained by Strang. For a survey of what is known
to date, see [G].
3 In particular, F is bounded (a result that dates back to a 1915 paper of Hadamard).
deal [with sth; see also: address, concern, consider, handle, take up, treat] To deal with Tf,
we note that..... [= Regarding Concerning) Tf, we note] Although [1] deals mainly with the
unit disc, most proofs are so constructed that they apply to more general situations. It became
clear that the Riemann integral should be replaced by some other type of integral, better suited
for dealing with limit processes. A survey of the research on f
n
(x, y) up to 1970 (most of it
dealing with the case n = 1) was given in [3]. Instead of dealing with lines one by one, we deal
with collections of lines simultaneously. This topic has been dealt with by many authors.
decay These estimates only require that f have a certain polynomial rate of decay at innity.
decide [see also: determine, settle, tell] Note that this lemma does not give a simple criterion
for deciding whether a given topology is indeed of the form T
f
. It becomes impracticable to
compute the zeros of F for degrees greater than 6; in any event, deciding whether the divisors
38 JERZY TRZECIAK
found in this way represent irreducible curves becomes increasingly dicult. In the event of a
tie, the winner is decided by the toss of a coin.
declare We partially order M by declaring X < Y to mean that.....
decline 1 [see also: decrease] We observe the decline in the number of unemployed.
2 [see also: decrease, diminish] The number of full-time sta has declined from 300 to just 50.
decompose [into sth; see also: split] If H is decomposed in the form H =..... Then F can be
decomposed according to the eigenspaces of P. Note that X decomposes into a direct sum of
2-planes.
decomposition a decomposition of X into irreducible components
decrease 1 [see also: decline] There is also a decrease in clustering as n increases, particularly
for normally distributed X. This decrease is oset by the contribution from the poles. Prots
in May show a modest decrease.
2 [see also: diminish, lower, drop, limit, reduce, cut down] We thus decrease P(n) by 2. After
each comparison the index is decreased. The left side of (3) obviously cannot decrease if r
increases.
dedicate [see also: devote] We can now formulate the problem to which the rest of this article
is dedicated.
deduce [see also: conclude, nd, infer, follow] To deduce (9) from (5), choose..... At this
stage we appeal to Theorem 2 to deduce that..... The case when f is decreasing can be proved
similarly, or else can be deduced from.....
deep [see also: profound] It is hoped that a deeper understanding of these residues will help
establish new results about the distribution of modular symbols.
default Throughout, all rings are by default k-algebras.
defer [see also: postpone] Making use of a technical estimate, the proof of which is deferred to
Section 3, we establish in Section 2 the estimates for..... We defer the proof of the moreover
statement in Theorem 5 until after the proof of the lemma.
dene [see also: declare, let, set] To dene P
n+1
from P
n
, we let..... We dene a the) function
f by setting f = Tg...... We dene the set T to consist of those f for which..... We dene a
map to be simple if..... Now M is dened to be the set of all sums of the form..... Here F
is only dened up to an additive constant. Here u
+
and u

are the positive and the negative

parts of u, as dened in Section 5. As dened in Section 1, these are structures of the form.....
The function f so dened satises..... The notion of backward complete is dened analogously
by exchanging the roles of f and f
1
. The fact that the number T(p) is uniquely dened,
even though p is not, enables us to dene the nullity of A as follows. The other two dening
properties of a -algebra are veried in the same manner.
deniteness [see also: idea, specic] Here Q can, for deniteness, be taken as Ff. These
extra stipulations are unimportant, but are given for deniteness.
denition We make the following provisional denition, which is neither general nor particularly
elegant, but is convenient for the induction which is to follow. With Lemma 2 in mind, we make
the following ad hoc denition. The preceding denitions can of course equally well be made
MATHEMATICAL ENGLISH USAGE 39
with any eld whatsoever in place of the complex eld. He proved the following theorem (see
Section 2 for pertinent denitions). The rst of the above equalities is a matter of denition,
and the second follows from (3). Another way is to extend the denition of the index to closed
curves by setting..... Our denition agrees with the one in [3]. The denition is legitimate
correct), because..... The precise denitions follow.
degree [see also: extent, level, stage] of degree at most k in some degree to in) a lesser
degree A number of authors have considered, in varying degrees of generality, the problem of
determining..... He received his master and PhD degrees from the University of Texas. Let
ABC be an angle of sixty degrees. The two lines intersect at an angle of ninety degrees. This
is the lattice packing rotated 45
o
. a 180
o
rotation
delete [see also: drop, omit, remove, dispense with] Note that there is a mistake on p. 3 of [5],
where the condition..... should be deleted. The quiver Q
1
is the same as Q but with x deleted.
deliberately [see also: intentionally, accidental] We deliberately write SP(m)rather than
P(m)in the complex case, because..... They dened the concept solely in terms of the norm
of the Banach space, deliberately avoiding any extra structure.
delicate [see also: subtle, dicult, ne] The two examples, E
1
and E
2
, dier by only a single
sequence, e, and they serve to illustrate the delicate nature of Theorem 2. To get around this
delicate issue, we shall separate the variables s and u.
demand 1 [see also: requirement, stipulation] The demand that each entry be a perfect square
results in nine equations. [Note the subjunctive be.]
2 [see also: require] Part 2 of the proof demands only that k > 1. A computational restraint
is the algebraic number theory involved in nding these ranks, which will typically be more
demanding than in our example of Section 1.
demonstrate [see also: prove, show, establish, verify, display, exhibit, indicate] The solution
is not in L(E) unless a stability condition is imposed, as was demonstrated in Example 3.
denote [see also: call, designate, write, set, symbol, label, relabel, rename] Denote by P the
space of..... [Not: Note P the space of..... or Denote P the space of.....] We let T denote
the set of..... .....with e
0
denoting multiplication by f. The wedge denotes that e
i
has been
omitted. Then we can nd a subsequence (still denoted by a
n
) such that a
n
< 1 for all n.
depart We depart from these previous works in our use of the non-ergodic versions of the basic
machinery. In this chapter we shall depart from the previous notation and use the letter m not
for Lebesgue measure, but for Lebesgue measure divided by (2)
1/2
.
depend [on sth; see also: base, rely, rest] However, the space Z depends crucially on the
particular ambient algebra A chosen. Note that O(g) depends on g only through its dierential
dg. We have to keep track of how the constant K depends on the domain D. We do not know
how V depends on the various choices made. The answer depends on how broadly or narrowly
the term matrix method is dened. Up to equivalence it only depends on the pair (f, g).
We divide them into three classes depending on the width of k. The proof of (2) is a matter
of straightforward computation, and depends on the relation ab cd = 1. These functions are
sometimes called elementary factors. Their utility depends on the fact that.....
dependence [see also: reliance] The holomorphic dependence of the integral on l follows from
40 JERZY TRZECIAK
(4). For convenience we ignore the dependence of f on g. For simplicity, we suppress the
explicit dependence on x in the notation. When there is no ambiguity we drop the dependence
on B and write just Y
T
for Y
T,B
. Strictly speaking, we should write something like a(l, m, n)
to reect the dependence; we shall rely upon context instead. We must now bring dependence
on d into the arguments of [9].
dependent An i-dependent lower bound is automatic by continuity.
depict [see also: illustrate] This graph is depicted in Fig. 1.
depth [see also: detail, expand, length] We now wish to discuss in some depth the problem
of.....
derivation Our approach provides an alternative derivation of the functional equation. For our
purposes here, the best way is to base the proof on the following theorem, a derivation of which
can be found in [P].
derive [see also: obtain, originate] This is derived in Section 3 along with a new proof of
Morgans theorem. In Section 2 the reader will be reminded of some important properties of
Bernoulli numbers, and some auxiliary results will be quoted or derived. However, we know
of no way of deriving one theory directly from the other. Here we consider a dual variational
formulation which can be derived similarly to that for the sandpile model. This term derives
from quiver, a notion used in representation theory of algebras. Most of the proofs in this
paper derive from those given by Fox for geometrically nite maps in [F]. The inequality (2.4)
is essentially contained in [LS] but will be rederived in Corollary 5 below.
describe [see also: account] Let F be as just described. As we let t vary, f(t) describes a curve
in M. It is interesting that (1) is a necessary condition in a much larger class of functions,
which we now describe. We describe how the notion of positivity relates to the other properties.
There are quite a number of cases, but they can be described reasonably systematically. the
process described
description [see also: account] In order to make our description of this process precise, we
dene..... [Not: to precise our description] We give a fairly simple description of a wide
class of averaging operators for which this rate of growth can be seen to be necessary. Our aim
here is to give some sort of functorial description of K in terms of G. The description of Q
makes it evident that (3) holds. [Note that the it is necessary here.] What is still lacking is an
explicit description of ker C.
deserve A formula like (3) surely deserves some explanation. As Corollary 2 shows, it is certainly
a question deserving further exploration. If nothing else, I hope to convince my readers that
Segals theorem deserves recognition as a profound contribution to Gaussian analysis.
design [see also: aim, intend] Condition (a) is so designed that a(x) = 0. The denition of
generator is designed to make the proof above work for M = Z.
designate [see also: denote, represent, call, name, term] This group will be designated by E
3
(not to be confounded with Euclidean 3-space). Each x here really designates the pair (x, Ax).
desirable However, if B were omitted in (1), the case n = 0 would imply Nf = 1, an undesirable
restriction.
MATHEMATICAL ENGLISH USAGE 41
desire 1 [see also: aim, motivation] The present paper is motivated by the desire to make the
subject as accessible as possible. It seems that the relations between these concepts emerge
most clearly when the setting is quite abstract, and this (rather than a desire for mere generality)
motivates our approach to the subject.
2 [see also: wish, want, require] Now F is the desired lifting. So F must be constant on H,
thereby showing that A = B, as desired claimed/required). Here L can be taken as large as
desired.
detail 1 [see also: depth, expand, fully, length, aspect, ingredient, feature, characteristic, point]
Unfortunately, the details of the calculations were omitted, and there is some doubt on whether
the result is correct since our analysis suggests that P
2
must vanish to third order; the presence
of L
2
is also suspect. In this section we discuss in some detail the relationship between.....
In more detail, the assertion is this: if..... To go into this in detail would take us too far aeld.
In detail the classication is complex, but in essence it is simple. For more details we refer
the reader to [4]. We refer the reader to the body of the paper for details.
2 [see also: extensive, thorough] In Section 4 we worked out a fairly detailed picture of linear
H-systems. A detailed exposition, more suited to the purposes of the present article, is given
in [9]. By detailed numerical calculation one may show that.....
deteriorate The estimate (3.5) deteriorates for xed N and increasing M.
determine [see also: decide, settle, tell] This normalization determines V uniquely. The
function f (initially dened on C
0
) determines a functional on S. ....., where G is uniquely
determined up to unitary equivalence up to an additive constant). Note that f is determined
only to within a set of measure zero. The order of G is completely determined by the assumption
that..... Over the past ten years the isomorphic structure of spaces of weighted holomorphic
functions has been largely [= almost completely] determined. We conclude that whether a
space X is an RG-space is not determined by the ring structure of C(X). The semigroup F
can be explicitly determined. Let us now take a quick look at the class N, with the purpose
of determining how much of Theorems 1 and 2 is true here.
develop [see also: build, construct, create, establish, work out, expand, grow] Kirk, building
on work of Penot, developed a more abstract version of..... Kearnes developed a commutator
theory for relative congruences, with the expectation that it can be used to prove Pigozzis
conjecture. This is most readily shown using the theory developed in Section 6. In this
subsection, we develop a formalism that will be used in the proofs of our main results.
development [see also: advance] Our development will require a detailed understanding of
cozero covers. We base our development on two properties of prolongation peculiar to this
case.
device [see also: method, means] This device makes it possible to replace multivalued functions
by functions with.....
devote [see also: dedicate] Much of the rest of the paper is devoted to a general study of.....
Most of this book is devoted to proving this theorem. [Not: devoted to prove]
diameter The support of F has diameter not exceeding l. a set with of) diameter 1
dier [see also: vary] Apart from a few embellishments necessitated by some technical dicul-
ties, the ideas dier very little from those used to prove Lemma 4. The two functions dier
42 JERZY TRZECIAK
at most on a set of measure zero. The distributions U and V dier only merely) by scale
factors from the distribution Z. We shall nd it convenient not to distinguish between two
such sequences which dier only by a string of zeros at the end. Thus F and G dier by an
arbitrarily small amount. The two codes dier only in the number of their entries. Then f(x)
and f(y) dier in at least n bits. We produce an evolution equation which diers from (2.3)
only in the replacement of the F
2
term by F
3
. If A = B, how does the situation dier from
the preceding one? This is the same as asking which row vectors in R have diering entries at
positions i and j.
dierence [see also: distinction, contrast] There is a marked dierence between X and Y .
One of the major dierences between F and G is that..... The dierence between these maps
is primarily in their kneading sequences. The main dierence from the case of nite coding
trees is the presence of limits. For size 1 this makes no dierence, but for sizes 2 and 3 it
considerably cuts down the number of cases to be considered.
dierent [from sth; amer. than sth; see also: distinct, other, various, several, particular, dis-
parate] The case a = 1 requires a dierent approach. The situation is quite dierent if we
replace H(U) by certain subclasses. by using a somewhat dierent method This idea is very
little dierent from what can already be found in [2]. Our solution is completely dierent from
theirs. [Not: much dierent] There is another, entirely dierent, way to see that A = B.
Namely, one can rst show that.....
dierently [see also: way] This terminology is used slightly dierently in [KA]. A further
complication arises from BP, which works rather dierently from the other labels.
dicult [see also: hard, complicated, complex, involved, intricate, delicate] When r > 3 things
become much more dicult. It becomes impracticable to compute the zeros of F for degrees
greater than 6; in any event, deciding whether the divisors found in this way represent irreducible
curves becomes increasingly dicult. This abstract theory is not in any way more dicult than
the special case of the real line.
diculty [see also: complication, problem, trouble] We now indicate how that diculty can be
circumvented. To get around overcome) this diculty, assume..... The proof of Theorem 6
bypasses this diculty. We now indicate some of the inherent diculties. But this obvious
attack runs into a serious diculty. The analogue of Theorem 1 presents no diculty. The
diculty disappears entirely if we think of our functions as elements of E. The only diculty
is in showing that..... As M is ordered, we have no diculty in assigning a meaning to (a, b).
[Not: diculty to assign] The diculty is that it is by no means clear what one should mean
by a normal family. The diculty consists in generalizing (b). Some such diculty is to be
expected. The assumption that the test statistics are identically distributed can be relaxed
without much diculty.
digit the tens digit the units digit
digress Let us digress to locate the need for relative ergodicity in the above arguments.
dimension A similar result holds in higher space dimension. in dimension n Leray and
Schauder laid the foundations for the generalization of the Lefschetz index to innite dimensions.
diminish [see also: decrease, decline, lower, reduce, cut down] If any element is removed from
an orthonormal set, its span is diminished. As n decreases, the class of such covers diminishes.
MATHEMATICAL ENGLISH USAGE 43
The dimension of a simplex is its cardinality diminished by 1.
direct [see also: straightforward] 1 Here is a simple direct proof. Tietzes theorem is a fairly
direct consequence of Urysohns lemma.
2 We direct the reader to [Aus] for a more detailed discussion of this problem.
direction For the other direction, take..... The if direction is trivial. We pause to record
a generalization of Theorem 2 in a dierent direction. Proceeding further in this direction, we
obtain the following corollary. Let A denote the rectangle B rotated through /6 in a clockwise
direction about the vertex (0, 1). The method of proof of Theorem B can be adapted to extend
the right-to-left direction of Mostowskis result by showing that..... the derivative of f at x in
the) direction v a direction pointing downward with respect to inequality in the opposite
direction
directly [see also: straight] Let us now prove directly (without recourse to [5]) that..... It
should be possible to prove this directly by studying the xed point set of the action of G.
They were dened directly by Lax [2], essentially as we have dened them. In the remainder of
this section, we study some properties of K, with the eventual aim (not realized yet) of describing
K directly using G.
disadvantage [see also: shortcoming, weakness, fail] Our method has the disadvantage of not
being intrinsic. A mild disadvantage of our quantile trees is that the prior to posterior com-
putation is not analytically tractable.
disappear [see also: vanish] The diculty disappears entirely if we think of our functions as
elements of E. One should remark that the ambiguity alluded to in Remark 3 disappears when
talking about an ane eld.
discover [see also: nd, learn] Thorin discovered the complex-variable proof of Rieszs theorem.
This strikingly simple proof was discovered by J. Dixon.
discuss [see also: consider] In this section we discuss in some detail the relationship..... This
is discussed more fully in [5]. These Banach algebras have been much discussed recently.
We put o discussing this problem to Section 5. Rather than discuss this in full generality,
let us look at a particular situation of this kind. the problem discussed [Not: the discussed
problem]
discussion [see also: consideration, examination, study, treatment] We now turn to a brief
discussion of another concept which is relevant to Johns theorem. We direct the reader to [Aus]
for a more detailed discussion of this problem. We summarize some of its main properties,
borrowing from the elegant discussion in Hensons article. Of course, it is tacitly understood
that it is this measure that is really under discussion. See [KT] for discussion of this technical
point. We have beneted greatly from discussions with many people, including..... a detailed
full/in-depth/brief/preliminary) discussion
disjoint [from sth; not: with sth] This means that all intervals we consider are either contained
in V or disjoint from V .
disparate [= clearly dierent] The approach in [GT] provides a unied way of treating a wide
variety of seemingly disparate examples.
44 JERZY TRZECIAK
dispense [with sth; see also: remove, drop, delete, omit] The following example shows that con-
dition (2) cannot generally be dispensed with if V is to be continuous. [= cannot be discarded]
display [see also: demonstrate, exhibit, show, indicate] Iterated correspondences display many
of the features of..... Table 1 displays the median quantile of the number of misclassied tumor
samples for each classier. the second displayed equation of the proof
disprove The conjecture will be disproved by exhibiting.....
disregard [see also: ignore, neglect] In the following, the errors of truncation will be disregarded.
distance [see also: away] the set of points with at) distance 1 from K the set of points at a
distance less than 1 from K Hence A and B are at distance precisely d. Consider a pair of
points a, b at distance 1. The point p is within distance d of X. Their centres are a distance
at least N apart. Any point not in B is moved by f a distance equal to twice the distance
to M. If we stay a xed distance o the critical line, we do not expect Benford behaviour.
distinct [see also: dierent] a sequence of distinct complex numbers The number of distinct
values that could be in a memory cell is at most s. However, only ve of these are distinct.
distinction [see also: dierence] The distinction between these two realizations lies in the
treatment of..... We shall proceed without making explicit distinctions between the two types
of convergence. There is only a ne distinction between the two schemes.
distinguish [see also: set apart] The Knaster continua are distinguished by the property that.....
Generally we add a tilde to distinguish between quantities associated with

G and those associ-
ated with G. We shall nd it convenient not to distinguish between two such sequences which
dier only by a string of zeros at the end. Where it is important to distinguish dierent norms
on E, we shall use the notation..... An arc is oriented if it has one of its endpoints distinguished
as the starting point and the other as the nishing point. Throughout the proof, E(n) (as
distinguished from E(n, A)) denotes the energy without the eld (5.1). Functions which are
equal almost everywhere are indistinguishable as far as integration is concerned.
distribute The error is multinomially distributed with mean zero and covariance matrix A.
There is also a decrease in clustering as n increases, particularly for normally distributed X.
distribution The random variable X has the Poisson distribution with mean v. Now X has a
normal N(0, 1) distribution. Let S
n
have the binomial distribution with parameters n and p.
It is intuitively clear that the amount by which S
n
exceeds zero should follow the exponential
distribution.
divide The proof will be divided into a sequence of lemmas. [Not: divided in a sequence]
Next divide J in half. On dividing through by f, we see that.....
do [see also: make] We shall do this by showing that..... This is done to simplify the notation.
As the space of Example 3 shows, complete regularity of X is not enough to allow us to do that.
For binary strings, the algorithm does not do quite as well. Recent improvements in the HL-
method enable us to do better than this. In fact, we can do even better, and prescribe nitely
many derivatives at each point of M. A geodesic which meets bM does so either transversally
or..... We have not required f to be compact, and we shall not do so except when explicitly
stated. This will hold for n > 1 if it does for n = 1. Consequently, A has all geodesics closed
if and only if B does. We can do a heuristic calculation to see what the generator of x
t
must
MATHEMATICAL ENGLISH USAGE 45
be. In contrast to the previous example, membership of D(A) does impose some restrictions
on f. We may (and do) assume that..... The statement does appear in [3] but there is a
simple gap in the sketch of proof supplied. Only for x = 1 does the limit exist. In particular,
for only nitely many k do we have F(a
k
) > 1. [Note the inversion after only.] In doing this
we will also benet from having the following notation.
document Since most of the results presented are quite classical (the novelty lies in the arrange-
ment, and some of the proofs are new), I have not attempted to document the source of every
item.
dominate [see also: bound, estimate] The right hand side is dominated by 2M.
doomed [= certain to fail; see also: futile] This attempt is doomed because the homogeneity
condition fails to hold.
double [see also: twice] The constant C may be taken as double the constant appearing in (5).
doubly We shall thus avoid doubly counting the same contribution.
doubt The only points (z, w) at which the continuity of g is possibly in doubt have z = 0.
Unfortunately, the details of the calculations were omitted, and there is some doubt on whether
the result is correct since our analysis suggests that P
2
must vanish to third order; the presence
of L
2
is also suspect.
down The numbers inserted in L must increase strictly down each column. This accords with the
intuition that as we pass down the coding tree, we nd out more and more detailed information
about the ordering actually represented. We show that..... by reverse induction on i, starting
at i = n and working down to i = 0. Thus, everything comes down to proving the existence
of M. For size 1 this makes no dierence, but for sizes 2 and 3 it considerably cuts down the
number of cases to be considered. Important analytic dierences appear when one writes down
precisely what is meant by.....
downward(s) We adopt the convention that the rst coordinate i increases as one goes down-
wards, and the second coordinate j increases as one goes from left to right. a direction pointing
downward with respect to [Note that downward and downwards can be used without dis-
tinction as adverbs, but the standard form of the adjective is downward (e.g., in a downward
direction).]
draft The author thanks H. Miller for a careful reading of an earlier draft. The author gratefully
acknowledges the referees helpful comments pertaining to the rst draft of this paper.
draw This gure is drawn to a scale of one to ten. The proof shows that if the points are drawn
at random from the uniform distribution, most choices satisfy the required bound. We shall
draw heavily on ideas from [3].
drawback A drawback to Polya trees, and perhaps the main reason why they have not seen
much application within the Bayesian nonparametric literature, is that an arbitrary partition
tree needs to be specied.
drop [see also: decrease, decline, reduction, omit, remove, delete, dispense with] Whenever the
dimension drops by 1, the rank drops by at most Z. The temperature dropped ten degrees.
For brevity, we drop the subscript t on h
t
. We drop the subscript when confusion is unlikely.
When there is no ambiguity we drop the dependence on B and write just Y
T
for Y
T,B
. We
46 JERZY TRZECIAK
show that one can drop an important hypothesis of the saddle point theorem without aecting
the result. Note that we can drop the factor of 1/2 by Lemma 3. This term drops out when
f is dierentiated.
due Observe that A is a vector space. This is due to two facts:..... We have F = G = H, the last
equality being due to the fact that..... Theorems 1 to 3 are classical results, due to Lebesgue.
This proof is due to R. P. Boas. [In the above examples, due is an adjective; the use of due to
as a preposition meaning because of, e.g. Due to Theorem 5, this map is bounded, although
quite common, is considered incorrect by some native speakers of English.]
duration For the duration of this proof only, let us write E
X
to denote.....
during We may require that the point P lie in one of the trees constructed before or during the
ith stage of the induction.
E
each [see also: all, any, every] Then F is bounded on each bounded set. Each of these three
integrals is nite. These curves arise from....., and each consists of..... There remain four
intervals of length 1/4 each. Here X assumes values 0, 1, ..., 9, each with probability 1/10.
Then F
1
, ..., F
n
vary each in the interval [0, 1]. The rst and third terms are each less than /3.
a progression each of whose terms can be written as..... These n disjoint boxes are translates
of each other. The two notions of rank are independent of each other.
ease [see also: simplicity] 1 For ease of notation, set I = I
f
. We note that the assumption of
GCH is made for convenience and ease of presentation.
2 To ease notation, we write..... Furthermore, adopting this strategy considerably eases con-
structing a coding tree from a linear order.
easily One easily checks that whenever a sequence a
n
obeys the uniform bound a
n
< C, one
has..... The series can easily be shown to converge. This can be easily reformulated in purely
geometric terms. However, as we are about to see, this complication is easily handled. The
passage from bounded f to general f is easily eected. It is easily seen that..... The hypotheses
of [4] are dierent, however, and do not seem to adapt easily to the time-inhomogeneous case.
easy [see also: simple, straightforward, elementary] A stronger topology makes it easier for a
given function to be continuous. It is almost as easy to nd an element such that..... The
other inequality is just as easy to prove. For these considerations it will be convenient to state
beforehand, for easy reference, the following variant of..... It is an easy matter to use Theorem
10 to construct all manner of interesting Peano continua [= continua of dierent kinds]
eect 1 [see also: consequence, result, outcome, inuence] This map will be shown to be in
M by examining its eect on B. We can immediately use one of these poles to cancel out
compensate for) the eect of the pole at s = 0. Osetting the eect of the pole at t = 0
requires more work. Conjecture 2 of [KH], to the eect that [= meaning that] there is no
relation P with E(P) = 1, still remains open.
2 We rst prove the (rather simpler) Theorem 7, by eecting a quite general reduction of the
problem to the study of certain isotropy factors. The passage from bounded f to general f is
easily eected.
eective [see also: successful] This provides an eective means for computing the index.
MATHEMATICAL ENGLISH USAGE 47
eectively It is obvious that the above theorem supplies an algorithm to eectively recognize
whether SP is in A.
eectiveness This follows immediately from (8) coupled with the fact that.....
eort [to do sth; at doing sth; of doing sth] It is also clear that there are extensions to.....,
but they do not seem to be worth the eort of formulating them separately. In an eort to
generalize Roberts method, we gave in [2] the following criterion for non-nite generation of
kernels. The rst of these was suggested by J. Serrin, who showed how to modify my earlier
treatment of..... so as to obtain stronger results with no extra eort.
either [see also: both] In either case, it is clear that..... [= In both cases] Now equate the
coecients of x
2
at either end of this chain of equalities. By Corollary 2, distinct 8-sets have
either zero, two or four elements in common. Each f can be expressed in either of the forms
(1) and (2). The two classes coincide if X is compact. In that case we write C(X) for either
of them. Either f or g must be bounded. Any map either has a xed point, or sends some
point to its antipode. But B is not divisible, hence C cannot be divisible either.
elaborate [see also: complicated, intricate] Zilber has a more elaborate denition of this notion,
but in fact his extra conditions are redundant. One reason for using this rather elaborate model
is that it permits a simple and concrete denition of the realization.
elegant We make the following provisional denition, which is neither general nor particularly
elegant, but is convenient for the induction which is to follow. We summarize some of its main
properties, borrowing from the elegant discussion in Hensons article.
element [see also: member, ingredient, feature, detail] We shall call the elements of such a
chain links. Computing f(y) can be done by enumerating A(y) and testing each element
for membership in C. Two consecutive elements do not belong both to A or both to B.
a 3-element set
elementary [see also: easy, simple, straightforward, basic, primary] It is an elementary check
that A is a vector space. However, shortly after learning about Wieners work, P. Levy found
a more elementary argument. By elementary algebra, we can show that.....
eliminate By repeated squaring to eliminate the radicals in this equality, we obtain.....
else The case when f is decreasing can be proved similarly, or else can be deduced from..... We
now exploit the relation (15) to see what else we can say about G. Little else is known about
the Klein property in this class of spaces. If nothing else, I hope to convince my readers that
Segals theorem deserves recognition as a profound contribution to Gaussian analysis.
elsewhere Dene A to be the matrix with 1 in the (i, j) entry and 0 elsewhere.
embed [or: imbed; in sth] The correspondence f Af embeds F as a closed ideal in G. The
next lemma shows how such a semilattice looks when embedded in a larger compact semilattice.
embedding [or: imbedding; of X in/into Y ]
embrace [see also: include, cover] This study embraces a number of interesting aspects of the
theory.
48 JERZY TRZECIAK
emerge [see also: arise, occur, result] It seems that the relations between these concepts emerge
[= become apparent] most clearly when the setting is quite abstract, and this (rather than a
desire for mere generality) motivates our approach to the subject.
emphasis [see also: stress] In this paper, emphasis is on the case where..... However, the
connection with Gromovs work has been obscured in recent years by an emphasis (in the
algebraic topology literature) on conguration spaces. We add the word positive for em-
phasis. In his Stony Brook lectures, he laid great emphasis on the use of..... considerable
increased/particular/special/strong/main) emphasis
emphasize [see also: stress, underline] When it is necessary to emphasize one particular coor-
dinate, we write..... Let us emphasize that K was not assumed to be connected in Theorem 4.
Our presentation is therefore organized in such a way that the analogies between the concepts
of topological space and continuous function, on the one hand, and of measurable space and
measurable function, on the other, are strongly emphasized.
employ [see also: apply, use, utilize] In discussing structures, we shall employ the standard
terminology of rst-order logic. The condition..... can be improved by employing a strategy
similar to that underlying the proof of Theorem 2. In this section we employ a version of the
Brown inequality to estimate N
k
.
enable [sth; sb to do sth; see also: allow, permit, possible] Repeated application of Lemma 2
enables us to write..... [Or: enables one to write; not: enables to write] Also, wherever
possible, we work with integer coecients, enabling us to obtain information about torsion.
In fact, there is a trick which enables one to reduce the time dependent case on R
n
to the time
independent case on R
n+1
. This enables discontinuous derivations to be built.
encircle The curve C encircles the origin twice.
encompass [see also: cover, include] Our notion of nonanalytic integral encompasses such well
known examples as..... We begin by extending Construction 2.1 to encompass B-algebras.
encounter [see also: meet, come across, run into] However, we immediately encounter the
problem of nonregularity of the data. We shall encounter similar situations again, and shall
apply convergence theorems to them without further comment.
encourage The results have been encouraging enough to merit further investigation. We en-
courage the reader unfamiliar with techniques from the theory of..... to consult [BS].
end 1 [see also: aim, purpose] At the end of Section 2, we prove..... [Not: In the end of]
Now F is dened to make G and H match up at the left end of I. We shall nd it convenient
not to distinguish between two such sequences which dier only by a string of zeros at the end.
To this end, we rst consider..... [= For this purpose; not: To this aim] Thus in the end
F will be homogeneous. [= nally]
2 [see also: conclude, nish, terminate] The path ends at x. The word ends in a. a word
starting with a and ending with b The exact sequence ends on the right with H(X). We end
this section by stating without proof an analogue of..... Arguing as before, we shall end up
with a simple tree all of whose facets contain V .
enhance [see also: improve, increase] It should be possible to enhance the above theorem
further by allowing an arbitrary locally compact group L. The usefulness and interest of this
MATHEMATICAL ENGLISH USAGE 49
correspondence will of course be enhanced if there is a way of returning from the transforms to
the functions, that is to say, if there is an inversion formula.
enjoy However, not every ring enjoys the stronger property of being bounded. On the other
hand, as yet, we have not taken advantage of the basic property enjoyed by S: it is a simplex.
enough [see also: suce, sucient, suciently] Choosing n small enough that na < 1 gives
the estimate..... This class is wide enough to include a number of examples of interest. The
preceding proof contains a result which is interesting enough to be stated separately. Put this
way, the question is not precise enough. It is therefore enough to show that..... It is also
enough to consider only real m. But this time boundedness on U is enough; we do not need
continuity on V . ....., which, by another theorem of Kimneys, is more than enough to guarantee
that P gives A outer measure 1. As the space of Example 3 shows, complete regularity of X
is not enough to allow us to do that. Repeating this procedure enough times gives.....
ensuing [= following] The idea of the ensuing computations is the following:.....
ensure [see also: guarantee] Theorem 3 is remarkable in that considerably fewer conditions than
in the previous theorems ensure universality. The hypothesis n > 1 ensures that Lemma 2 is
applicable.
entail [see also: involve, necessitate, require, necessarily] The analogous construction when A
is not ane entails the following subterfuge. [= requires the following trick] This entails that
some smoothness assumptions will have to be made.
enter Here G enters essentially in an algebraic way. If the line segment L does not enter the
interior of A, then neither does L

.
entire [see also: whole, complete, full, total] Since the entire argument is based solely upon
assumption (6.1), the conclusion of the theorem must hold.
entirely [see also: completely, fully, wholly] The diculty disappears entirely if we think of our
functions as elements of E. Each component which meets X lies entirely within Y . There is
another, entirely dierent, way to see that A = B. Namely, one can rst show that.....
entry Dene A to be the matrix with 1 in the (i, j) entry and 0 elsewhere. Take A to be the
matrix with all entries zero except for i j at (i, j).
enumerate [see also: list] There is not space to enumerate them all here. Computing f(y)
can be done by enumerating A(y) and testing each element for membership in C.
enumeration Let r
1
, r
2
, . . . be an enumeration of the rationals in [0, 1].
envisage Even though more general probabilistic constructions could be envisaged [= contem-
plated], we focus on those schemes where.....
equal 1 Then A equals B. [= A is equal to B; not: A is equal B] The degree of P equals
that of Q. Now J is dened to equal Af, the function f being as in (3). [= where the function
f is.....] Therefore this condition is equivalent to the correlation equalling 1 B.
2 The resulting metric space consists precisely of the Lebesgue integrable functions, provided
we identify any two that are equal almost everywhere.
50 JERZY TRZECIAK
equality Equality holds occurs) in (9) if..... ....., with equality if a = 1. There is equality if
a = 1. Equality is achieved only for a = 1. Then a b c a. We therefore have equality
throughout. The relation becomes an equality if the w
i
form an orthonormal basis.
equally [see also: just] But H itself can equally well be a member of S. The method applies
equally well to certain other error terms. All inputs of size n are equally likely to occur.
However, as observed in [5], the proof given in [7] is equally valid for regular sets.
equate Equating coecients we see that..... Equating the coecient of x
2
in V to zero, we
get..... Now equate the coecients of x
2
at either end of this chain of equalities.
equation Equation (2) now reads Ax =..... We can solve the resulting equations successively
for c
1
, ..., c
n
. [Do not overuse equation for expressions without equality sign; use estimate,
inequality, formula, relation etc. instead.]
equip The algebra D(F) comes equipped with a dierential d such that H(d) = 0.
equivalent [to sth; to doing sth] This is equivalent to requiring saying) that..... Therefore
this condition is equivalent to the correlation equalling 1 B. The following three statements
are equivalent:..... For this it is sucient to check that x is in V , or, what is equivalent, that
a(x) is bounded. Many equivalents are known for this property.
equivalently Thus F can be equivalently dened as F =..... Then f = g, or equivalently
a(f) = a(g). Firstly, if G is abelian, does it follow that S(G) = K, or equivalently, that
X
G
= p(M(G))?
erroneous [see also: fallacious, invalid] Our main nding in this paper is that this intuition
turns out to be erroneous.
error [see also: mistake] The equations are satised with error at most O(n). The saddle-point
conditions are satised up to an error o(n). We take this opportunity to correct a minor error
in Lemma 2 of [PS]. Unfortunately, there is a simple but serious error in the nal step of the
proof.
especially [see also: notably, particularly, particular] This may appear rather wasteful, espe-
cially when n is close to m, but these terms only give a small contribution to our sum. An
especially interesting case occurs when.....
essence [see also: core, substance, nature, point, heart] In detail the classication is complex,
but in essence it is simple.
essential [see also: basic, critical, crucial, key] Essential to the proof are certain topological
properties of G. The assumed positivity of u
n
is essential for these results. We shall not use
this fact in any essential way.
essentially [see also: basically, largely, mostly, mainly] The operator P satises essentially
the same inequality as does F. Even in the case n = 2, the application of Theorem 6 gives
essentially nothing better than the inequality..... Here G enters essentially in an algebraic way.
They were dened directly by Lax [2], essentially as we have dened them.
establish [see also: prove, show, demonstrate, set] We need only establish (8) for intervals of the
form..... They established the Hasse principle subject to a rank condition on the coecients.
To establish uniqueness, suppose..... This establishes (i). The problem now reduces to
MATHEMATICAL ENGLISH USAGE 51
establishing that..... One might hope that in the particular case of the GL energy, this could
be established, but we do not see an easy path to such a conclusion.
establishment We now work toward the establishment of properties (A) to (D).
estimate 1 We now give an estimate for..... in terms of..... The weight satises a weak type
(1, 1) estimate. a crude rough/sharp/precise) estimate
2 [see also: bound, dominate, assess] To estimate F from above below), consider.....
estimation [see also: estimate] Theorem 3 can be applied to the estimation of b
k
. The
remainder of the estimation is largely as before with B replaced by C. [Note that estimation
is an activity; its result is an estimate.]
even Actually, the proof gives an even more precise conclusion:..... Nevertheless, it might be
possible to make sense of (2) even for non-injective V by considering a multi-valued operator Z.
This is true even if H
2
(B) = 0, as evidenced by the existence of manifolds whose ane base
is a Klein bottle. Even though we were able to derive a formula for....., it is not easy to use.
The fact that the number T(p) is uniquely dened, even though p is not, enables us to dene
the nullity of A as follows.
event [see also: case] It becomes impracticable to compute the zeros of F for degrees greater
than 6; in any event, deciding whether the divisors found in this way represent irreducible curves
becomes increasingly dicult. In the event of a tie, the winner is decided by the toss of a coin.
eventual [= occurring at the end; ,= possible; see also: ultimate, nal] It is impossible to predict
the eventual outcome of the process. The eventual aim is to obtain a closed form expression
for F. In the remainder of this section, we study some properties of K, with the eventual aim
(not realized yet) of describing K directly using G.
eventually [= in the end; ,= possibly] The iterates eventually reach the value 1. Then we can
nd some net (s
k
) which eventually leaves every compact subset of G. It will eventually appear
that the results are much more satisfactory than one might expect. an eventually increasing
sequence
ever The question of whether B is ever strictly larger than A remains open.
every [see also: all, any, each, arbitrary] To every F there corresponds a unique..... For every
g in X not in X) there exists an X..... [But: for all f and g, for any two maps f and g; every
is followed by a singular noun.] Every invariant subspace is of the form..... [Do not write:
Every subspace is not of the form..... if you mean: No subspace is of the form.....; every has to
be followed by an armative statement.] However, not every ring enjoys the stronger property
of being bounded. Every possible such sequence gives rise to..... The generally accepted point
of view in this domain of science seems to be changing every few years.
evidence 1 [see also: proof, data, material, fact] There is no evidence to the contrary. Com-
puter evidence suggests the dynamics of these maps is rich and varied. Some partial evidence to
support this conjecture is discussed in [3]. We oer numerical evidence to support a conjecture
that there exist innitely many primes of this type.
2 [see also: show, prove] This is true even if H
2
(B) = 0, as evidenced by the existence of
manifolds whose ane base is a Klein bottle.
52 JERZY TRZECIAK
evident [see also: clear, obvious, plain] From this previous work, it was not at all evident that
there would be some covering or non-covering properties which would be preserved by one of
these kinds of extensions. The description of Q makes it evident that (3) holds. [Note that the
it is necessary here.]
evidently [see also: clearly, manifestly, obviously] The algebraic properties (i) and (ii) required
of j are evidently true.
exactly [see also: precisely] Actually, [3, Theorem 2] does not apply exactly as stated, but its
proof does. The function F vanishes to order exactly n at zero.
examination [see also: check, consideration, discussion, inspection, scrutiny, verication] An
examination of the argument just given reveals that this is all we have used. Examination of
the left and right members of (1) shows that.....
examine [see also: study, explore, investigate, look] Examining how the Lipschitz constant
depends on F, we nd that..... The algorithm examines only roughly one-quarter to one-third
of the characters. By carefully examining the relations between the quantities U
i
, we see that.....
example [see also: illustration, exemplify] To see an example, set..... As an example, let.....
Here is an example of..... An explicit example is two planes through the origin. This can
obviously happen for some X: examples are c
0
and l
1
. We can make this clear with the following
example. The simplest example of this is furnished by..... We now describe a more interesting
example featuring an operator that is non-trivially supercyclic. It is easy to see, by means of
an example, that..... Show by an example that a P-system need not be a eld. Examples are
given to show that..... Examples abound in which P is discontinuous.
exceed [see also: outnumber, surpass] The degree of F exceeds that of G by at least 4. The
support of F has diameter not exceeding l. It is intuitively clear that the amount by which S
n
exceeds zero should follow the exponential distribution. Dene a
k
to be the probability that
exactly k out of the 2n values X
i
exceed T, conditional on X
0
> T.
except [see also: apart from, besides] 1 for all n except a nite number for no x except the
unique solution of..... [Note the dierence between besides, except and apart from: besides
usually adds something, except subtracts, and apart from can be used in both senses; after
no, nothing etc., all three can be used.] Then G is dierentiable except for a jump at x = 0.
Except for these two lemmas, we make no use of the results of [4]. That isexcept the use
of relaxed controlsprecisely the stochastic Bellman equation. Then F is invertible except
possibly on an at most countable set. Thus, F is invertible except where at) x = 0. Hence
F is invertible except at countably many points. The desingularization of f is X except when
Q and Q

have a common factor, in which case..... We have not required f to be....., and we
shall not do so except when explicitly stated. The assumptions of Section 3 are still in force
except that for greater generality we do not assume f to be continuous.
2 The trivial cases when the family consists of only one set are excepted.
exception There are few exceptions to this rule. As the proof will show, these properties, with
the exception of (c), also hold for complex measures. With the exception noted below, we
follow Stanleys presentation [3, Sec. 2].
exceptional Only in exceptional circumstances is it true that f(x +y) = f(x) +f(y). [Note the
inversion.]
MATHEMATICAL ENGLISH USAGE 53
exercise It is an elementary exercise to show that.....
exchange [see also: interchange] The notion of backward complete is dened analogously by
exchanging the roles of f and f
1
. It is highly likely that if one of the Xs is exchanged for
another, the inequality fails.
exclude [see also: preclude, prevent, rule out] The condition A < B excludes quite a few of the
standard Young functions. It remains to exclude the case where..... Note that we explicitly
exclude from the values of a simple function. The possibility A = is not excluded.
exclusive The two cases are not mutually exclusive.
exclusively We will work exclusively in the category of standard Borel probability spaces, and
so will often suppress mention of their sigma-algebras.
exemplify [see also: example, witness] The elasticity of M is clearly at least equal to n (as
exemplied by the element x).
exhibit [see also: demonstrate, reveal, show, display, indicate, express] Formula (7) exhibits
the same phenomenon. The resulting formula exhibits u as the Laplace transform of x.
This rather counterintuitive phenomenon is exhibited by the following example. A function
exhibiting this type of behaviour has been constructed by Cox [7].
exist Note that dimE, if it exists, is..... There exists a function f and a constant c such that.....
[Or: There exist a function f and a constant c] The limit lim
x0
f(x) exists. [Not: There
exists a limit lim
x0
f(x).] What relations exist between A and B?
existence [see also: presence] We rst show that existence of a solution of..... provides a
sucient condition for optimality. Given f and w, we shall prove the existence of a solution
to (Q). This result shows that the mere existence of a nontrivial automorphism j of M produces
the cut I(j) of M that satises (2).
expand [see also: develop, work out, detail, length] Theorem 3 below expands on this idea.
Expand f in powers of x. The purpose of this paper is to expand substantially the class of
maps for which the index can be computed.
expansion The Laurent expansion of f around about) zero is..... expansion in powers of x
the base p expansion representation) of x
expect [see also: hope] Hence we would expect the functions..... to behave similarly. We need
to check that F-derivatives behave in the way we expect with regard to sums, scalar multiples
and products. Compact multipliers, as one would expect, are those elements of A which.....
It seems reasonable to expect that....., but we have no proof of this. That is the least one
can expect. There is no reason to expect this to be an inverse map on K, but we do have
the following. However, F is only nonnegative rather than strictly positive, as one may have
expected. It will eventually appear that the results are much more satisfactory than one might
expect. Some such diculty is to be expected. Along the way, we come across some perhaps
unexpected rigidity properties of familiar spaces.
expectation [see also: hope] Kearnes developed a commutator theory for relative congruences,
with the expectation that it can be used to prove Pigozzis conjecture.
54 JERZY TRZECIAK
expense [see also: cost, price, sacrice] At the expense of replacing b by b
2
we may remove the
condition.....
explain [see also: describe, account, reason, justify] This explains why we chose 9 rather than,
say, 1 for the second coordinate. In Chapter 5, we shall explain what it means for a subset V
of A to be determining for the centre of X. To obtain the required map one must modify the
method as explained in [9]. Any other unexplained notation is as found in Fox (1995).
explanation [see also: answer, solution, explication] The explanation for this denition is
that..... This corollary calls for some explanation and comment. A formula like (3) surely
deserves some explanation. a lucid clear/plausible/likely/straightforward) explanation
explication A complete explication of the Fox spaces is warranted, as it will likely reveal further
clues to the dierences between the parabolic and hyperbolic theories.
explicit We write K(Q) for K to make the dependence on Q explicit. This module is denoted
by H(X), or H(X, R) if we want to make explicit the coecient ring. For simplicity, we
suppress the explicit dependence on x in the notation. We shall proceed without making
explicit distinctions between the two types of convergence. Throughout what follows, we shall
freely use without explicit mention the elementary fact that..... An explicit example is two
planes through the origin. Here is a more explicit statement of what the theorem asserts.
explicitly [not: explicitely] Explicitly, we have the formula..... We have not required f to
be....., and we shall not do so except when explicitly stated. The semigroup F can be explicitly
determined. Note that we explicitly exclude from the values of a simple function.
exploit [= make full use of] We now exploit the relation (15) to see what else we can say
about G. Our rst result generalizes (8) by exploiting some general facts seemingly overlooked
by the aforementioned authors. This connection has been exploited on occasion [= sometimes]
to construct various innite families of regular maps.
exploration [see also: analysis, investigation] As Corollary 2 shows, it is certainly a question
deserving further exploration.
explore [see also: examine, investigate, study] The question of..... has been explored under a
variety of conditions on A.
expose [,= present] [Do not write We expose examples of maps if you mean: We present
examples of maps; to expose something means to uncover it (if it has been hidden) or make
it known (if it has been secret).]
exposition [see also: presentation] The proof is mainly included to keep the exposition as self-
contained as possible. A detailed exposition, more suited to the purposes of the present article,
is given in [9]. For convenience of exposition, we work with an error term of the form..... For
completeness of exposition, we now recall the denition of the ordering on R. The exposition
owes much to the work of Dold described in [3]. The author thanks the referee for recommending
various improvements in exposition.
express [see also: show, demonstrate, indicate, convey] We can express f in terms of..... Each
f can be expressed in either of the forms (1) and (2). The chart shows government spending
expressed as a proportion of national income. This is sometimes expressed by saying that.....
We would like to express our thanks to all of these institutions for their hospitality and support.
MATHEMATICAL ENGLISH USAGE 55
expression We shall abbreviate the expression (3) to F(k). For this choice of , and with
u = z = s, the expression (5.3) simplies greatly. The eventual aim is to obtain a closed form
expression for F. How many such expressions are there? the expression in brackets in braces,
in parentheses) the rightmost expression [= the last one on the right] the right-hand side
expression
extend [see also: continue, generalize] We can extend f by zero to the whole . Now A, B
and C all extend to a small neighbourhood of x. We extend f to be homogeneous of degree 1.
We make D a Poisson algebra by extending the Poisson bracket on A by linearity. We begin
by extending Construction 2.1 to encompass B-algebras. This procedure can be extended to
take care of any number of terms. This conclusion extends to the general diraction problem.
Much of the foregoing can be extended to the noncompact case. The method of proof of
Theorem B can be adapted to extend the right-to-left direction of Mostowskis result by showing
that.....
extension [see also: generalization] Then F has no holomorphic extension to any larger region.
[BS] contains an extension of Proposition 2 to the setting of nitely additive set functions.
an extension of f o U
extensive [see also: thorough, large, wide, substantial, detail] There is now an extensive lit-
erature dealing with..... For a comprehensive treatment and for references to the extensive
literature on the subject one may refer to the book [M] by Markov. An extensive treatment of
the h-principle can be found in [6].
extensively [see also: fully, detail, expand, length] This subject has recently been extensively
studied.
extent [see also: scope, range, degree, far] It is not clear to what extent this can be generalized
to other varieties of loops. The aim of this article is to study the relationship between the size
of A, as measured by its diameter, and the extent to which A fails to be convex. In this section
we ask about the extent to which F is invertible.
extra [see also: additional, further, more] There is an extra statement that causes a new char-
acter to be read. We nd ourselves forced to introduce an extra assumption. These extra
stipulations are unimportant, but are given for deniteness. The method sketched in Section
3 of [Con] carries through with our choice of =
1
+
2
, but there is one extra ingredient
worthy of mention. The rst of these was suggested by J. Serrin, who showed how to modify
my earlier treatment of..... so as to obtain stronger results with no extra eort. They dened
the concept solely in terms of the norm of the Banach space, deliberately avoiding any extra
structure.
extract We extract the following from the proof of Lemma 5 of [KL].
extreme 1 Two natural extremes arise that one has to consider: the case when..... and the case
when..... At the other extreme, L could be empty. At the other extreme is the set of tent
maps with a dense orbit.
2 An extreme example is when F is a singleton.
56 JERZY TRZECIAK
F
fact [see also: information, detail, evidence, proof, actually, more] A stronger result is in fact
true. We note that H is in fact not Lipschitz continuous if this condition is violated. In fact,
we can do even better, and prescribe nitely many derivatives at each point of A. Now, just
the fact that F is a homeomorphism lets us prove that..... The advantage of applying..... lies
in the fact that..... Observe that (1) just uses the fact that m is unary.
factor 1 E is similar to F but scaled by a factor 1/3. Thus E is similar to F with a scale factor
1/3. The only additional feature is the appearance of a factor of 2. Thus we need only alter
our constants by a factor of 2 to deal with this case. This accounts for the factor 1/2 in (4).
2 The map f factors through the space X. We can factor g into a product of irreducible
elements.
fail [see also: hold, invalid, short] The space X fails to have the Radon-Nikodym property.
Hence A fails to satisfy (3). For j = 1 the operator is bounded, yet the integral (8) fails to be
nite. It is highly likely that if one of the Xs is exchanged for another, the inequality fails.
fairly [used with positive properties; see also: rather, reasonably] Tietzes theorem is a fairly
direct consequence of Urysohns lemma. To see that f = g is fairly easy. In the next theorem,
we give fairly minimal conditions that imply..... There are, however, a few important papers
of which we were unaware until fairly recently.
fall The most frequently used models fall into one of the following two categories. This example
falls within the scope of Coxs theorem. The method falls short of providing an explicit formula
for the index. [= fails to provide] In the latter case we may simply adjust F to equal 1 on the
Borel set where it falls outside the specied interval.
fallacious [see also: invalid, erroneous] However, this argument is fallacious, because as re-
marked after Lemma 3,.....
false The theorem is denitely false without the assumption that....., as an inspection of Exam-
ple 3 shows. It is this proposition that we believe to be false in Morava E-theory. Suppose,
towards a contradiction, that this is false. This works regardless of whether B is true or false.
There are nite measures for which the omission of (iv) would make (b) false.
familiar [see also: common, customary, usual, standard] The reader is assumed to be familiar
with elementary K-theory. Let E be Cantors familiar middle thirds set. We encourage the
reader unfamiliar with techniques from the theory of..... to consult [BS].
family [see also: collection, set] Let I be the family of all subalgebras which contain F. [Or:
that contain F; you can use either that or which in dening clauses.] The diculty is that it
is by no means clear what one should mean by a normal family. The words collection, family
and class will be used synonymously with set.
far [see also: extent, much] This problem is still far from being solved. Its description is far
from complete. To go into this in detail would take us too far aeld. However, (1) has been
proved so far only in the case where..... So far we have not topologized M(R). So far it seems
not to be known whether the geometric condition on X can be omitted. However, we have
thus far been unable to nd any magic squares with seven square entries. Functions which are
equal almost everywhere are indistinguishable as far as integration is concerned.
MATHEMATICAL ENGLISH USAGE 57
fashion [see also: manner, way, method] This was proved in elementary fashion by P. J. Cohen.
We shall need ways of constructing new triangulations from old ones which alter the f-vector
in a predictable fashion. Continuing in this fashion, we get a collection V
r
of open sets, one
for every rational r, with..... in a similar fashion
fast [see also: rapidly] a fast decreasing function [Not: a fastly decreasing function] The
other player is one-third as fast.
favorable In (3.6), the most favorable case is when t is in the centre of the interval.
feasible [see also: possible] It does not appear feasible to adapt the methods of this paper to.....
feature [see also: characteristic, detail, element, ingredient] 1 One unusual feature of the solu-
tion should be pointed out. The remarkable feature of this theorem is that..... The only addi-
tional feature is the appearance of a factor of 2. The main new feature is the use of the face ring
to produce lower bounds for the number of vertices. This t has the feature we want. Iterated
correspondences display many of the features of..... a basic distinctive/essential/key/main/
major/signicant/dominant/special) feature
2 We now describe a more interesting example featuring an operator that is non-trivially super-
cyclic. This article features results in both spectral theory and operator ergodic theory made
possible by a recent renewal of interest in the consequences of Jamess inequalities.
fellowship The paper was commenced whilst the second author held a Fullbright Fellowship.
few [see also: some, several, less] There are few exceptions to this rule. [= not many] There are
few, if any, other signicant classes of processes for which such precise information is available.
Few of various existing proofs are constructive. He accounts for all the major achievements in
topology over the last few years. The generally accepted point of view in this domain of science
seems to be changing every few years. This set has no fewer elements than K has. [Not: no
less elements; less should not be followed by a plural countable noun. However, use less when
it is followed by than or when it appears after a noun: X has no less than twenty elements; Y
has ten elements or less.] Any vector with three or fewer 1s in the last twelve places has at
least eight 1s in all. Theorem 3 is remarkable in that considerably fewer conditions than in
the previous theorems ensure universality. Therefore, F has the fewest points when the index
vanishes. There are a few exceptions to this rule. [= some] Apart from a few embellishments
necessitated by some technical diculties, the ideas dier very little from those used to prove
Lemma 4. Many interesting examples are known. We now describe a few of these. Only a
few of those results have been published before. Quite a few of them are now widely used.
[= A considerable number] The condition A < B excludes quite a few of the standard Young
functions.
eld [see also: area] Further motivation for looking at ideal class groups comes from the eld
of cryptography. his eld of scientic interest
gure As shown in Figure 3, neither curve intersects X. These patterns are illustrated in Fig. 4.
An example is plotted in Figure 1. The graph of Fig. 5 shows that..... This gure is drawn
to a scale of one to ten.
ll The reader is encouraged to ll in the missing details. Sets of the form (6) are disjoint and
ll out any such orbit.
58 JERZY TRZECIAK
nal [see also: eventual, ultimate] This paper is in nal form. In the nal section of the paper,
we list some open problems.
nally [see also: last, lastly] Finally, multiplication by a permutation matrix will get the expo-
nents in descending order. With Lemma 4 in at) hand, we can nally dene E to be equal to
P(m)/H. This nally yields f = g. [Not: yields that f = g]
nd [see also: come across, discover, see, obtain] Continuing in this manner, we see that for
4, the curve C
4k+3
is found in the half-strip B. The main results of the paper are found in
Section 2. Our proof of Theorem 2 is based upon ideas found in [BN]. Any other unexplained
notation is as found in Fox (1995). Following the same lines we nd that it takes k prolongations
to get an immersed curve. This will help us nd what conditions on A are needed for T(A)
to be analytic. Then, for any two xed points that Wagners method does not nd to be
equivalent, he considers the possible lengths of potential solutions to (1). We nd ourselves
forced to introduce an extra assumption. This accords with the intuition that as we pass down
the coding tree, we nd out more and more detailed information about the ordering actually
represented. Our main nding in this paper is that this intuition turns out to be erroneous.
ne [see also: delicate, subtle, minor] There is only a ne distinction between the two schemes.
A change in perspective allows us to gain not only more general, but also ner results than
in [ST].
nish [sth; doing sth; see also: close, complete, conclude, end, terminate] This nishes the
proof. Two more triangulations will nish the job. After having nished proving (2), we shall
return to..... [Not: nished to prove (2)] We nish by mentioning that, suitably modied,
the results of Section 2 apply to the AP case. Continuity then nishes o the argument.
rst 1 [see also: original] He was the rst to propose a complete theory of triple intersections.
Because N. Wiener is recognized as the rst to have constructed such a measure, the measure
is often called the Wiener measure. Let S
i
be the rst of the remaining S
j
. The rst two are
simpler than the third. [Or: the third one; not: The rst two ones] As a rst step we shall
bound A below. We do this in the rst section, which the reader may skip on a rst reading.
At rst glance, this appears to be a strange denition. The rst and third terms in (5) combine
to give..... the rst author = the rst-named author
2 [see also: initially, originally, beginning, rstly] First, we prove (2). [Not: At rst] We
rst prove a reduced form of the theorem. Suppose rst that..... His method of proof was to
rst exhibit a map..... In Lemma 6.1, the independence of F from V is surprising at rst. It
might seem at rst that the only obstacle is the fact that the group is not compact. [Note the
dierence between rst and at rst: rst refers to something that precedes everything else in a
series, while at rst [= initially] implies a contrast with what happens later.]
rstly [see also: rst] Firstly, if G is abelian, does it follow that S(G) = K, or equivalently,
that X
G
= p(M(G))?
t [see also: suit, suitable, satisfactory, appropriate] Other types t into this pattern as well.
We aim at tting this result into the general scheme of..... Although these proofs run along
similar lines, there are subtle adjustments necessary to t the argument to each new situation.
The key part is to show that the submanifolds U
k
t together to form a complex submanifold.
x For xed k,..... We temporarily x n in N. Fix n for the moment. Fix n and let c vary.
MATHEMATICAL ENGLISH USAGE 59
The mapping f leaves keeps) the origin xed. Thus F has x as its unique xed point.
focus 1 In this paper, the main focus is on the Riemann moduli space. Our focus now will be
on one-sided averages.
2 In this section we shall focus attention on..... Even though more general probabilistic con-
structions could be envisaged [= contemplated], we focus on those schemes where.....
-fold Then P covers M twofold. Sales increased almost fourfold in this period. The motivation
for writing this paper was twofold. By k-fold integration by parts,.....
folklore The following easy lemma is surely folklore. We now prove a simple fact about semi-
groups: this is surely a folklore result. The following simple example has long been a part of
ergodic-theoretic folklore.
follow [see also: imply, ensuing, subsequent] The remarks following Theorem 2 show that a = 0.
The analysis to follow only covers the case d = 1. We make the following provisional denition,
which is neither general nor particularly elegant, but is convenient for the induction which is
to follow. The following three statements are equivalent:..... For D a smooth domain, the
following are equivalent. The following has an almost identical proof to that of Lemma 2.
The idea of the ensuing computations is the following:..... His argument is as follows. The
fact that the number T(p) is uniquely dened, even though p is not, enables us to dene the
nullity of A as follows. In what follows In all that follows), L stands for..... Throughout
what follows, we shall freely use without explicit mention the elementary fact that..... It follows
that a is positive. [= Hence Consequently,/Therefore,) a is positive.] Firstly, if G is abelian,
does it follow that S(G) = K, or equivalently, that X
G
= p(M(G))? If we prove that G > 0,
the assertion follows. The general case follows by changing x to x a. We follow Kato [3]
in assuming f to be upper semicontinuous. Following the same lines we nd that it takes k
prolongations to get an immersed curve. The proof follows very closely the proof of (2), except
for the appearance of the factor x
2
. It is intuitively clear that the amount by which S
n
exceeds
zero should follow the exponential distribution.
for 1 We write z = (x, y) for the common point of A and B. We rst consider the M/G/1 queue,
where M (for Markov) means that..... For D a smooth domain, the following are equivalent.
For m not an integer, the norm can be dened by interpolation. For (ii), consider..... [= To
prove (ii), consider] Now (3) is clear. As for (4), it is an immediate consequence of Lemma 6.
[= Concerning (4)] Thus F is integrable for the product measure. Then for such a map to
exist, we must have H(M) = 0. The problem with this approach is that V has to be C
1
for
(3) to be well dened. Computing f(y) can be done by enumerating A(y) and testing each
element for membership in C. Therefore, the system (5) has a solution of the sought-for type.
2 [see also: because] We must have Lf = 0, for otherwise we can replace f by fLf. [= because
otherwise] It turns out that it suces to show that A = 1, for if this is proved, the preceding
remark shows that..... [This use of for sometimes leads to confusion; e.g., never write: for
x X if you mean: since x X. Also, avoid starting a sentence with a For in this sense.]
[Do not use for followed by an ing-form to indicate the purpose of ones actions: instead of: For
showing that f is bounded, we observe, write: To show that f is bounded, we observe. However,
you can use for to indicate the purpose of a thing: This provides a method for recognizing pure
injective modules.]
force [see also: strength] 1 The next lemma is the rst in this section that seems to require the
60 JERZY TRZECIAK
full force of the normality of X. The assumptions of Section 3 are still in force except that for
greater generality we do not assume f to be continuous.
2 [see also: make, necessitate, impose, require] If we choose close to 1, we can force r < 0.
This forces f to satisfy (6). We nd ourselves forced to introduce an extra assumption.
foregoing [see also: above, precede] How much of the foregoing can be extended to the non-
compact case?
form 1 [see also: shape, structure, type, kind] The map F can be put brought) into this form
by setting..... Let S be the set of all solutions of (8) of the form (3). We shall then show that
this f can be represented in the form f =..... This implies that the local martingale must take
a very specic form. in diagonal form
2 [see also: constitute, make up, account for, represent] Consider the Blaschke product formed
with the zeros of f. They form a base of the topology of X. Theorem 2 will form the basis
for our subsequent results. The key part is to show that the submanifolds U
k
t together to
form a complex submanifold.
formalism In this subsection, we develop a formalism that will be used in the proofs of our main
results.
formally The sum in (2), though formally innite, is therefore actually nite.
former [see also: latter] Then either....., or..... In the former case,..... Examples 1 and 2 give
two operators, the former bounded and the latter not, with..... The denition is stated in terms
of local martingales, rather than martingales, for the technical reason that the former are easier
to characterize in applications.
formula [pl. formulas or formulae] The main information conveyed by this formula is that.....
These models furnish integral formulas for the matrix entries of F. It seems appropriate to
mention in passing the corresponding formula for the cohomology of B
n
. In addition to illus-
trating how our formulas work in practice, it provides a counterexample to Browns conjecture.
Find integral formulas by means of which the coecients c
n
can be computed from f. For
later use, we record the following formulas:.....
formulate [see also: state] We can now formulate the problem to which the rest of this article is
dedicated. It is also clear that there are extensions to....., but they do not seem to be worth the
eort of formulating them separately. The benet of formulating our notion of isomorphism
section as above will become clear shortly.
formulation [see also: statement] In the next section we introduce yet another formulation of
the problem. Here we consider a dual variational formulation which can be derived similarly
to that for the sandpile model. It is proved in [1] (albeit with a slightly dierent formulation)
that.....
forthcoming In this section we gather some miscellaneous results that are more or less standard.
These will be used to calculate the constants c

, and we place them here to avoid interrupting

the forthcoming arguments.
fortunately Fortunately, there is a very satisfactory solution to this problem, due to Vermes.
forward If t does not appear in P at all, we can jump forward n places.
MATHEMATICAL ENGLISH USAGE 61
foundation [see also: basis, underlie] In Section 2, we lay the foundations for a systematic
study of..... Leray and Schauder laid the foundations for the generalization of the Lefschetz
index to innite dimensions.
fraction [see also: proportion, ratio] A mere fraction of this energy is currently utilized.
framework [see also: context, setting, set-up] This is actually a special case of the preceding
framework. [Not: preceeding]
free Then H is a free R-module on as many generators as there are path components of X.
freedom It is the freedom of choice of D in this construction that enables us to.....
freely Throughout what follows, we shall freely use without explicit mention the elementary fact
that.....
frequently [see also: often, repeatedly] Part (c) is a frequently used criterion for the measura-
bility of a real-valued function. We shall frequently write w.w. for weakly wandering. This
sort of proof will recur frequently in what follows.
from This is immediate from 3.2. We see from (2.3) that..... From (2.3) we have..... Then X
is the Swiss cheese obtained from the family D. Thus A can be written as a sum of functions
built up from B, C, and D. From now on, F will be xed. Consider the family of ordered
triples of elements from F. The main dierence from the case of nite coding trees is the
presence of limits. Clearly, the contribution from those r with A(r) > 0 can be neglected.
In Lemma 6.1, the independence of F from V is surprising at rst. The presence here of the
direct summand H is simply to prevent A from having disconnected spectrum.
full [or: fulll] We now full the promise made at the start of the proof by handling the case
of G = S
n
.
full [see also: complete, entire, whole, total] For a fuller discussion of this topic, see [6].
fully [see also: completely, entirely, detail, expand, extensively, length] However, to our knowl-
edge this is not fully resolved. This is discussed more fully in [5].
functor The assignment of K
1
to K, and of T
1
to T, denes a functor between the category of
commutative algebras and the category of compact semigroups with continuous homomorphisms.
furnish [see also: give, oer, provide, supply, aord, yield] These models furnish integral for-
mulas for the matrix entries of F. Then Lemma 2 furnishes the bound f ..... The simplest
example of this is furnished by.....
further [see also: additional, extra, more] Thus every subsequence contains a further subse-
quence converging weakly to some limit. The only case requiring further analysis occurs when
f = 0. The results have been encouraging enough to merit further investigation. As Corollary
2 shows, it is certainly a question deserving further exploration. We omit further details.
A further complication arises from BP, which works rather dierently from the other labels.
We shall encounter similar situations again, and shall apply convergence theorems to them with-
out further comment. Proceeding further in this direction, we obtain the following corollary.
It should be possible to enhance the above theorem further by allowing an arbitrary locally
compact group L.
62 JERZY TRZECIAK
furthermore [see also: also, moreover, likewise] Furthermore, adopting this strategy consider-
ably eases constructing a coding tree from a linear order.
futile [see also: doomed] Having established (1), one might be tempted to try to extend this
result to general p through the choice of a suitable ideal B. Alas, as we shall see now, this
attempt is futile.
future For future reference, we record this in the following corollary. We quote for future
reference another result of Fox: there exists..... This is an interesting area for future research.
For future use, choose any monotone h(m) tending to innity such that..... These upper
bounds are too large to be useful in computer calculations in general, but the ideas in the proofs
will surely contribute to better bounds in the future.
G
gain [see also: get, obtain, achieve] The theorem gains in interest if we realize that..... Thus it
is reasonable to attempt, using this homeomorphism, to gain an understanding of the structure
of M. It is useful to consider some rather simple examples to gain some intuition. A change
in perspective allows us to gain not only more general, but also ner results than in [ST].
gap The statement does appear in [3] but there is a simple gap in the sketch of proof supplied.
The theory of correspondences may be viewed as bridging the gap between.....
gather [see also: collect, combine, piece together] We gather here various notation for future
reference. In this section we gather some miscellaneous results that are more or less standard.
The data were gathered for about a year.
general His techniques work just as well for general v. How are these two optimality notions
related? In general, they are not. While topological measures resemble Borel measures, they
in general need not be subadditive. We suspect that more can be said in general about which
subextensions can be obtained from nite-rank modules, but we will not explore this matter
further here.
generality There is no loss of generality in assuming that..... This involves no loss of generality.
Without loss of generality we can assume that..... [In many cases, the phrase without loss
of generality can be omitted: write simply: We can clearly assume that..... Avoid using the
abbreviation w.l.o.g.] Without losing any generality, we could have restricted our denition
of integration to integrals over all of X. [Not: Without loosing] It simplies the argument,
and causes no loss of generality, to assume..... A completely dierent method was used to
establish Theorem 2 in full generality. Rather than discuss this in full generality, let us look at
a particular situation of this kind. A number of authors have considered, in varying degrees of
generality, the problem of determining..... It seems preferable, for claritys sake, not to present
the construction at the outset in the greatest generality possible. It seems that the relations
between these concepts emerge most clearly when the setting is quite abstract, and this (rather
than a desire for mere generality) motivates our approach to the subject.
generalization [see also: extension] Leray and Schauder laid the foundations for the general-
ization of the Lefschetz index to innite dimensions. We pause to record a generalization of
Theorem 2 in a dierent direction. It is not immediately obvious what this generalization has
to be.
MATHEMATICAL ENGLISH USAGE 63
generalize [see also: extend] Corollary 2 generalizes and strengthens Theorem 3 of [9]. This
approach does not seem to generalize to arbitrary substructures. It is not clear to what extent
this can be generalized to other varieties of loops.
generally However, M is generally not a manifold. It is not generally possible to restrict f to
the class D. More generally this argument also applies to characterizing Hurewicz subsets of I.
generate This is slightly at odds [= inconsistent] with the terminology of [4], as Fox denes the
trace lter to be the normal lter generated by A. The family of 4-sets will be used to generate
a symmetry outside N but in M.
get [see also: obtain, acquire, make, arrange, overcome, circumvent] Apply Theorem 3 to get
a function..... We thus get f = g. [Not: We thus get that f = g.] We do not expect to
get F closed. The induced topology is not compact, but we can always get it to be contained
in a Bohr topology. Finally, multiplication by a permutation matrix will get the exponents in
descending order. Fortunately, F does not get too close to p. To get around this diculty,
assume that..... It is also tempting to get round this problem by working with.....
give [see also: provide, oer, supply, furnish, aord, dene, yield, imply, present, produce] The
action of G is given by gf =..... The argument just given shows that..... This gives (1)
and shows that..... However, a slight strengthening of the hypotheses does give us a regular
measure. We now give some applications of Theorem 3. We give X the topology of uniform
convergence on compact subsets of I. Note that E can be given a complex structure by
setting..... Suppose we are given an f of the form f =..... Given > 0, we can nd such
that..... This is certainly reasonable for Algorithm 3, given its simple loop structure. He used
a new version of an algorithm for nding all normal subgroups of up to a given index in a nitely
presented group. A function given on G gives rise to an invariant function on G

.
glance [see also: look, sight] At rst glance Lemma 2 seems to yield four possible outcomes.
Neighbourhoods of points in these spaces appear at rst glance to have a nice regular struc-
ture, but upon closer scrutiny, one sees that many neighbourhoods contain collections of arcs
hopelessly folded up. Now, for arbitrary n, a glance at the derivative shows that.....
glue Then M is obtained by glueing gluing) X to Y along Z. There is a natural way to glue
the associated varieties together along their common boundary. The set A is obtained from B
by removing a neighbourhood of C and gluing in a copy of D.
go 1 [see also: stroke] Its role is to rule out having two or more consecutive P-moves (on the
grounds that they can be performed in one go).
2 [see also: continue, proceed, pursue, turn] a path obtained by going from A to B along the
lower half of the circle Here the interesting questions are not about individual examples, but
about the asymptotic behaviour of the set of examples as one or another of the invariants (such
as the genus) goes to innity. This proves that the dimension of S does not go below q.
We adopt the convention that the rst coordinate i increases as one goes downwards, and the
second coordinate j increases as one goes from left to right. Some members go into more than
one V
k
. To go into this in detail would take us too far aeld. We now go through the clauses
of Denition 3. Before going to the proof, it is worth noting that..... This idea goes back at
least as far as [3]. This argument goes back to Banach. Many of these results are known, and
indeed they go back to the seminal paper of Dixmier [D] of 1951. Going back to the existential
step of the proof, suppose that..... Before we go on, we need a few facts about the spaces L
p
.
64 JERZY TRZECIAK
The equation PK = 0 then goes over to QK = 0. The rest of the proof goes through as for
Corollary 2, with hardly any changes. There are kneading sequences for which the arguments
of Section 4 go through routinely.
good [see also: satisfactory, appropriate, suitable] The kernel satises good large time bounds
if.....
grant We claim that F is the product of two disjoint cycles. Granting this, we reach a contra-
diction as follows.
grasp [see also: understand, recognize, realize, see] To grasp the dierence between the two
notions, consider.....
grateful [not: greatful] The author is grateful to the referee for a number of helpful suggestions
for improvement in the article.
gratefully The author gratefully acknowledges the referees helpful comments pertaining to the
rst draft of this paper.
great [see also: large, more] Then F is 3 greater than G. Thus F is not no) greater than G.
Consequently, F is greater by a half. However, F can be as great as 16. One should take great
care with..... In his Stony Brook lectures, he laid great emphasis on the use of..... Another
topic of great interest is how much of adjunction theory holds for ample vector bundles. We
could have adopted an approach to proving Theorem 2 along a line of reasoning which bears
greater resemblance to the treatment of the analogous result in Section 1. To show the greater
simplicity of our method over Browns, let us..... It seems preferable, for claritys sake, not to
present the construction at the outset in the greatest generality possible.
greatly [see also: considerably, signicantly, substantially] For this choice of , and with
u = z = s, the expression (5.3) simplies greatly.
ground [see also: reason] Its role is to rule out having two or more consecutive P-moves (on
the grounds that they can be performed in one go).
grow [see also: develop, increase] The product being considered grows like n
3
. Our study grew
out of some valuable conversations with Kirk Douglas.
growth The structure of a Banach algebra is frequently reected in the growth properties of its
analytic semigroups. We give a fairly simple description of a wide class of averaging operators
for which this rate of growth can be seen to be necessary. the growth rate of V
n
as n
a function of moderate growth
guarantee [see also: ensure] Observe how the completeness of L
2
was used to guarantee the
existence of f. However, (5) is sucient to guarantee invertibility in A. We are guaranteed
only one dense product for each k. This guarantees that f satises all our requirements. .....,
which, by another theorem of Kimneys, is more than enough to guarantee that P gives A outer
measure 1.
H
half 1 We sketch the proof of the easy half of the theorem. This proves one half of (2); the
other half is a matter of direct computation. the left half of the interval the upper half of
the unit disc Thus F is greater by a half.
MATHEMATICAL ENGLISH USAGE 65
2 F is half the sum of the positive roots. [Or: half of the sum] Half of the sets of R miss i and
half the remaining miss j. Then J contains an interval of half its length in which f is positive.
On average, about half the list will be tested. Now F is half as long as G. [Or: as G is]
We divide N in half. The length of F is thus reduced by half.
hand Our presentation is therefore organized in such a way that the analogies between the
concepts of topological space and continuous function, on the one hand, and of measurable
space and measurable function, on the other, are strongly emphasized. On the other hand, F
fails to have property P. [Not: On the other side] With Lemma 4 in at) hand, we can
nally dene E to be equal to P(m)/H. Once the dissipation relation is in hand, no further
work is required. It can be easily calculated by hand that the nonzero solutions are.....
handle [see also: consider, deal, take up, treat, manipulate, control] The theorem indicates that
arbitrary multipliers are much harder to handle than those in M(A). But (9) needs handling
with greater care. However, as we are about to see, this complication is easily handled.
The map G can be handled in much the same way. This is handled by a direct case-by-case
argument.
happen [see also: occur] If this happens on a set of positive measure, then f cannot be contin-
uous. It may happen that B
1
is the only compatible compactication. What happens if the
word nonnegative is omitted? Under what conditions does it happen that r < s?
hard [see also: dicult, complex, complicated, involved, intricate] This is the hard part of
Joness theorem. The theorem indicates that arbitrary multipliers are much harder to handle
than those in M(A). The calculation of M(f) is usually no harder than the calculation of
N(f). The cases p = 1 and p = 2 will be the ones of interest to us, but the general case is no
harder to prove.
hardly [see also: little] The rest of the proof goes through as for Corollary 2, with hardly any
changes. [= with almost no changes] On the whole, the solution can hardly be considered
satisfactory. [= is rather unsatisfactory]
harm [see also: loss] The hypothesis f(0) ,= 0 causes no harm in applications, for if.....
have [see also: possession] This square has a perimeter equal to the circumference of the circle.
Then M is a Banach algebra having for its identity the unit point mass at 0. Thus R
has rank 2 determinant zero/cardinality c). Therefore F has a countable spectrum a nite
norm/a compact support). [Or: F has countable spectrum etc.] Since....., we have Tf equal
to 0 or 2. Then L consists of those g which have Kg(n) = 0 for all n. Is it possible to have
m(E) < 1 for such a set? However, X does have nite uniform dimension. Then B does
not have the Radon-Nikodym property. Let f be a map with f[M having the Mittag-Leer
property. Suppose A is maximal with respect to having connected preimage. This allows
proving the representation formula without having to integrate over X. It has to be assumed
that..... The problem with this approach is that V has to be C
1
for (3) to be well dened.
heart [see also: core, essence] This simple observation lies at the heart of the proof of our rst
theorem.
heavily [= a lot] Standard Banach space notation is used throughout. For clarity, however, we
record the notation that is used most heavily. We shall draw heavily on ideas from [3].
66 JERZY TRZECIAK
heavy [see also: cumbersome] This section makes heavy use of a theorem of Alsen and related
results.
help [see also: aid] This will help us nd what conditions on A are needed for T(A) to be
analytic. [Or: help us to nd] The knowledge of the invariant subspaces of an operator helps
us to visualize its action. It is hoped that a deeper understanding of these residues will help
establish new results about the distribution of modular symbols. To calculate (2), it helps to
visualize the S
n
as the successive positions in a random walk. It may help to think of F as
being a smooth approximation to the Heaviside function.
helpful [see also: useful, advantageous, valuable] It is helpful to keep these similarities in mind.
hence [see also: consequently, so, thus, therefore, follow] Then F is continuous, hence bounded
on D. We can suppose R is semilocal (and hence a PID).
henceforth [= hereafter, from this time on] This theorem will henceforth be referred to as the
minimum principle. [Note the double r in referred.] We henceforth identify SC(K K) with
a -subalgebra of L

(X X). Henceforth, therefore, we may assume that f is irreducible.

here Here is a simple direct proof. Here is a restatement:..... Here are some elementary
properties of these concepts. Here are some other situations in which we can draw conclusions
only almost everywhere. It is here assumed that..... Here and throughout, P(E) denotes.....
The parameter interval was here taken to be (0, 1).
hereafter [= henceforth, from now on] This theorem will hereafter be referred to as the minimum
principle. [Note the double r in referred.] This property will be used repeatedly hereafter.
heuristic We can do a heuristic calculation to see what the generator of x
t
must be.
high A similar result holds in higher space dimension. The vector eld H always points towards
the higher A-level.
highly It is highly likely that if one of the Xs is exchanged for another, the inequality fails. It
is generally a highly nontrivial question whether..... a highly informative book
hinder [= make sth dicult] There are other problems with this example which would hinder
any attempt to follow the proof given here too closely.
hold [see also: apply, true, valid, force] All our estimates hold without this restriction. The
desired inequality holds trivially whenever A > 0. In this section we investigate under what
conditions the converse holds. An obvious question to ask is whether the assertion of Theorem 1
continues to hold for..... Equality holds occurs) in (9) if..... This attempt is doomed because
the homogeneity condition fails to hold. [= The attempt is certain to fail] The rst author
holds a Rockefeller Foundation fellowship. He held the Courant Chair at New York University
for three years before his retiring. In the year 2000 In 2000), two important number theory
conferences were held at Princeton University.
hope 1 This leaves the hope that a ratio theorem may persist in a more general setting.
2 [see also: expect, expectation] A simple argument shows that we cannot hope to have Df = 0.
We cannot hope to say anything about the structure of each isotropy factor as a system in
its own right. Almost everywhere convergence is the best we can hope for. Specically, one
might hope that a clever application of something like Choquets theorem would yield the de-
sired conclusion. If nothing else, I hope to convince my readers that Segals theorem deserves
MATHEMATICAL ENGLISH USAGE 67
recognition as a profound contribution to Gaussian analysis. It is hoped that a deeper un-
derstanding of these residues will help establish new results about the distribution of modular
symbols.
hopeless [see also: impossible, impracticable] a hopeless task
hopelessly Neighbourhoods of points in these spaces appear at rst glance to have a nice regular
structure, but upon closer scrutiny, one sees that many neighbourhoods contain collections of
arcs hopelessly folded up.
hospitality We would like to express our thanks to all of these institutions for their hospitality
and support. This research was initiated when the rst author was visiting the University
of Alberta in the summer of 2008; the nancial support and kind hospitality are gratefully
acknowledged.
how We have to keep track of how the constant K depends on the domain D. A natural question
is how sharp the bounds given in Theorem 6 are. Another topic of great interest is how much
of adjunction theory holds for ample vector bundles. Observe how the completeness of L
2
was used to guarantee the existence of f. How many of them are convex? How many such
expressions are there? How many entries are there in this section? How many multiplications
are done on average?
however [see also: but, though, nevertheless, matter] 1 In this section, however, we shall not
use it explicitly. That approach was used earlier in [2]. There, however, it was applied in simply
connected regions only. [Avoid using however as a simple substitute for but.]
2 This implies that however we choose the points y
i
, the intersection point will be their limit
point. [= no matter how we choose] However small a neighbourhood of x we take, the image
will be..... Then M
n
(x) = 0 for every x, however large.
hypothesis [pl. hypotheses; see also: assumption, conjecture] We show that one can drop an
important hypothesis of the saddle point theorem without aecting the result. The hypothesis
f(0) ,= 0 causes no harm in applications, for if..... Assume, in addition to the hypotheses of
Exercise 4, that..... We now bound v on under) the hypothesis H
m
. In fact, we shall prove
our result under the weaker hypothesis that W is weakly bounded, rather than just bounded, on
an innite subset of G. However, a slight strengthening of the hypotheses does give us a regular
measure. We establish our results both unconditionally and on the assumption of the Riemann
Hypothesis. In geometric language, the hypothesis is that F is.....; part of the conclusion is
that F satises..... the hypothesis of positivity [= the positivity hypothesis] the hypothesis
of F being projective
I
idea [see also: concept, notion, objective, deniteness] Some of the basic ideas from functional
analysis are also included. The idea behind our use of the map is that..... The idea is to
relax the constraint of being a weight function in Theorem 3. It is this idea that underlies some
of the results of [2]. This idea goes back at least as far as [3]. Apart from a few embellishments
necessitated by some technical diculties, the ideas dier very little from those used to prove
Lemma 4. Our proof of Theorem 2 is based upon ideas found in [BN]. The proof makes use
of many of the ideas of the general case, but in a simpler setting. Although the idea is simple,
its implementation is complicated by the fact that..... It is a pleasure to thank R. Greenberg
68 JERZY TRZECIAK
for bringing his criterion for..... to our attention, and for generously sharing his ideas about
it. We shall draw heavily on ideas from [3]. To x ideas, assume that M is a pentagon.
a stimulating clever/original/alternative/basic) idea
identical [see also: same] The obvious rearrangement reveals the right side to be identical
with (8). The following has an almost identical proof to that of Lemma 2.
identify As a rst step we identify the image of . Letting m identies this limit as
H. Using the standard inner product we can identify H with H

. The tangent space to

N at x is identied with M via left translation. We henceforth identify SC(K K) with a
-subalgebra of L

(X X). The resulting metric space consists precisely of the Lebesgue

integrable functions, provided we identify any two that are equal almost everywhere.
identity Clearly, M is a Banach algebra having for its identity the unit point mass at 0.
i.e. [= that is; see also: namely] These are precisely the linear functionals which also preserve
multiplication, i.e., f(ab) = f(a)f(b). [The comma after i.e. is optional.] [Note the dierence
between namely and that is: while namely introduces specic or extra information, that is (or
i.e.) introduces another way of putting what has already been said.]
if [see also: when, whenever, whether, provided] The map F is said to be proper if G is dense.
[Note: no comma before if here.] By deleting the intervals containing x, if any, we obtain.....
There are few, if any, other signicant classes of processes for which such precise information
is available. The only thing to check is that maps compose correctly; but this is an easy, if
tedious, verication. We can assume, by decreasing n if necessary, that.....
ignore [see also: disregard, neglect] For convenience we ignore the dependence of f on g. The
reader may wonder why we have apparently ignored the possibility of obtaining a better lower
bound by considering..... [Do not write: We ignore if A exists if you mean: We do not know
if A exists.]
illegitimate [see also: justify] In the preceding proof, the appeal to the dominated convergence
theorem may seem to be illegitimate since.....
illuminate These three results lead to several illuminating pieces of information about the (in-
suciently studied) Berger property in general spaces.
illustrate [see also: show, demonstrate, exhibit, display] To illustrate, let us state the following
corollary. These patterns are illustrated in Fig. 4. In addition to illustrating how our formulas
work in practice, it provides a counterexample to Browns conjecture. The two examples, E
1
and E
2
, dier by only a single sequence, e, and they serve to illustrate the delicate nature of
Theorem 2. The following result illustrates the utility of (3).
illustration [see also: example] By way of illustration, here is an example of..... We can now
pose a problem whose solution will aord an illustration of how (5) can be used.
image the image of A under f = the f-image of A the inverse image = the preimage
immediate The rst conclusion is immediate from Theorem 1. Now (3) is clear. As for (4), it
is an immediate consequence of Lemma 6. [= Concerning (4)]
immediately However, we immediately encounter the problem of nonregularity of the data.
It is not immediately obvious what this generalization has to be. We can immediately use
MATHEMATICAL ENGLISH USAGE 69
one of these poles to cancel out compensate for) the eect of the pole at s = 0. This follows
immediately from (8) coupled with the fact that.....
implement This procedure, once implemented, can thereafter be applied with great eectiveness.
implementation Although the idea is simple, its implementation is complicated by the fact
that.....
implication [see also: consequence] The implication one way follows from Theorem 2. None
of the one-way implications in (1.1) can be reversed. The rigidity inequality has a strong
implication when the fundamental group of P is nontrivial. the converse implication = the
converse
implicit [see also: imply] The following lemma, crucial to Theorem 2, is also implicit in [4].
This is essentially implicit in [BC] but we cannot quite quote the result we need. The constants
implicit in the symbol depend on r.
imply [see also: follow, entail, implicit] Then x = y and y = z implies x = z. [Or: imply]
Note that M being cyclic implies F is cyclic. However, if B were omitted in (1), the case n = 0
would imply Nf = 1, an undesirable restriction. Our present assumption implies that the last
inequality in (8) must actually be an equality. That (2) implies (1) is contained in the proof
of Theorem 1 of [4]. The continuity of f implies that of g. We maintain the convention that
implied constants depend only on n. The equivalence of (a) and (b) is trivially implied by the
denition of M.
importance [see also: role, signicance, relevance, interest] Another group of importance in
physics is SL
2
(R). The next two theorems reveal the importance of this concept. of consider-
able critical/crucial/fundamental/outstanding/particular/primary/increasing/minor/secondary)
importance
important [see also: main, crucial, principal, major, signicant] It is important to notice some
of the weaknesses inherent in the above approach. Where it is important to distinguish dierent
norms on E, we shall use the notation..... We show that one can drop an important hypothesis
of the saddle point theorem without aecting the result. Important analytic dierences appear
when one writes down precisely what is meant by..... It has some basic properties in common
with another most important class of functions, namely, the continuous ones. These extra
stipulations are unimportant, but are given for deniteness.
impose [see also: force] Theorem 7 imposes a quantitative restriction on the location of the zeros
of..... This theorem removes the restriction to convex regions which was imposed in Theorem 8.
impossible [see also: impracticable, hopeless] It is impossible to predict the eventual outcome of
the process. ....., which is impossible. [Not: what is impossible] But....., it being impossible
to make A and B intersect. [= since it is impossible to make]
impracticable [see also: impossible, hopeless] It becomes impracticable to compute the zeros
of F for degrees greater than 6; in any event, deciding whether the divisors found in this way
represent irreducible curves becomes increasingly dicult.
improve [see also: enhance, rene, sharpen, progress, better] The following example shows that
the degree of smoothness predicted by Theorem 6 cannot be improved on. The condition.....
70 JERZY TRZECIAK
can be improved by employing a strategy similar to that underlying the proof of Theorem 2.
We note in passing that Fox has subsequently improved Barness result by showing that.....
improvement [see also: progress, renement] We now show the following improvement on (2).
The basic improvement is the smoothness condition on G given in (6). The object of this
paper is to obtain improvements in two cases, namely for forms of degree 7 and 11. There
has since been a series of improvements, of which we briey mention the work of Levinson.
The author is grateful to the referee for a number of helpful suggestions for improvement in the
article. The author thanks the referee for recommending various improvements in exposition.
a considerable major/marked/radical/remarkable/signicant/minor/slight) improvement
impulse [see also: inspiration, motivation] An original impulse for this investigation came from
the study of.....
in We put b in R unless a is already in. This equation has a solution in integers for N > 7.
Expand f in powers of x. It is this point of view which is close to that used in C

-algebras.
The maps f
t
are homotopic, so f
1
and f
0
induce the same maps in homology. These intervals
are disjoint from all those used in dening J
1
. In doing this we will also benet from having
the following notation. This also resolves the ambiguity introduced earlier in choosing an order
of the lifts of U. Each A
i
meets A in a nite set. [= A
i
A is nite] Then one Y
i
can
intersect another only in one point. Values computed for the right side of (2) were rounded
up in the fourth decimal place. Then F varies smoothly in t. Thus F
n
(x, y) converges to
F(x, y) uniformly in x. Clearly, A
j
is increasing in j. The word ends in a. in diagonal form
in geometric language In less precise language, the requirement is that the two angles are
the same in size and in orientation. Then P is the product of several integer factors of about
x
n
in size. The set A is roughly triangular in shape. The proof proper [= The actual proof]
will consist of establishing the following statements in sequence. convergence in probability in
distribution) Less than 1 in p of its points will result in a quartic with ideal class number p.
Theorem 3 is remarkable in that considerably fewer conditions than in the previous theorems
ensure universality. As M is ordered, we have no diculty in assigning a meaning to (a, b).
The prime 2 is anomalous in this respect, in that the only edge from 2 passes through 3. This
is where the notion of an upper gradient comes in. The set A is obtained from B by removing
a neighbourhood of C and gluing in a copy of D.
inability A shortcoming of our method is the inability to compare three or more progressions.
incidentally [see also: by-product, pass, way] Incidentally, our computation shows that.....
Incidentally, the question of whether K(E) is amenable for specic Banach spaces E seems to
have received almost no attention in the literature.
include [see also: embrace, cover, subsume, comprise, contain] We shall only use (2), but
have included (1) for completeness. Some of the basic ideas from functional analysis are also
included. The proof is mainly included to keep the exposition as self-contained as possible.
The nal lemma is due to F. Black and is included with his kind permission. We begin by
describing the class of functions f considered, which includes the special cases quoted above.
This paper enlarges the class of continua with this property, namely to include those which.....
Let Q denote the set of positive denite forms (including imprimitive ones, if there are any).
Let K and L be quasivarieties such that K is properly included in L. from stage A up to,
but not including, stage B
MATHEMATICAL ENGLISH USAGE 71
incomparable [with sth, to sth] By genericity of P there is an x incomparable with both y
and z.
incompatible [with sth] However, rigidity is incompatible with the existence of an absolutely
continuous component.
incomplete From the standpoint of linear programming, the above discussion is incomplete in
that it throws no light upon the question whether the function F attains its inmum.
increase 1 an increase of 5%a 5% increase) in the cost of living A small percentage reduction in
the cost of materials resulted in a signicant increase in prot. a considerable major/marked/
signicant/substantial/modest/small/sharp/steady/sixfold/overall) increase
2 [see also: enhance, grow, raise] Obviously, G
n
increases to G as n . Sales increased
almost fourfold in this period. The budget has increased by more than a third. We adopt the
convention that the rst coordinate i increases as one goes downwards, and the second coordinate
j increases as one goes from left to right. The index increases by 1 when the path is crossed
from right to left. The numbers inserted in L must increase strictly down each column. Also,
F does not increase distances. There has recently been increasing interest in the theory of.....
It becomes impracticable to compute the zeros of F for degrees greater than 6; in any event,
deciding whether the divisors found in this way represent irreducible curves becomes increasingly
dicult. Indeed, as n increases, it becomes increasingly rare for a manifold to be a hyperplane
section of another projective manifold.
indebt I am greatly indebted to S. Brown for this example. We are indebted here to Villanis
account (see [2]) of a standard generalization of convex conjugacy.
indeed Note that this lemma does not give a simple criterion for deciding whether a given
topology is indeed of the form T
f
. We now prove..... Indeed, suppose otherwise. Then..... We
leave to the reader the proof that f is indeed self-adjoint, and not merely symmetric. Many of
these results are known, and indeed they go back to the seminal paper of Dixmier [D] of 1951.
indenitely Suppose that the process continues indenitely.
independence [of sth from sth] In Lemma 6.1, the independence of F from V is surprising at
rst.
independent [of sth, not: on sth] ....., the constant C being independent of n. Note that
F is independent of the choice of the family S. Indeed, it is routine to verify that the index
so constructed is independent of the choices made. It turns out that this is independent of the
representations taken (as long as they are faithful). The two notions of rank are independent
of each other. a result of independent interest
independently Similar results were obtained independently in [AB]. All our results hold in-
dependently of whether the underlying eld is R or C. Here Q
j
for j = 1, ..., n are drawn
independently, conditional on the values generated at level m.
indeterminate We write V [[h]] for the space of formal power series in an indeterminate h.
indicate [see also: demonstrate, show, point, express, convey, suggest, mark] ....., where the
prime indicates that only terms with p > 0 may appear. We shall write a.e. [] whenever
clarity requires that the measure be indicated. We now indicate some of the inherent diculties.
72 JERZY TRZECIAK
The proof will only be indicated briey. We now show that G is in the symbol class indicated.
for k in the indicated range
indication The proof is nonconstructive and gives no indication of what the exceptional set may
look like.
indistinguishable Functions which are equal almost everywhere are indistinguishable as far as
integration is concerned.
individual [see also: particular, single, specic, unique] Here the interesting questions are not
about individual examples, but about the asymptotic behaviour of the set of examples as one or
another of the invariants (such as the genus) goes to innity.
individually Although individually these systems can still be quite complicated, a possibly more
tractable task is to describe their possible joint distributions.
induce The maps f
t
are homotopic, so f
1
and f
0
induce the same maps in homology. The
induced homomorphism is multiplication by 2. On TK we set up [= introduce] the symplectic
structure induced by the metric.
induction The proof is by induction on n. By decreasing induction on p,..... We show that.....
by reverse induction on i, starting at i = n and working down to i = 0. We proceed by
induction. If we apply induction to (3) we see that..... For the base step of the induction,
consider a vertex t in A. We may require that the point P lie in one of the trees constructed
before or during the ith stage of the induction. [Note the subjunctive lie.] Induction shows
that if.....
inductive This example indicates that the inductive argument in the proof of Theorem 1 is
unavoidable.
inductively Thanks to Lemma 2, we can now modify the proof of Theorem 3 by inductively
constructing ve sequences..... Dene two polynomials f and g inductively as follows.
inequality opposite reverse) inequality inequality in the opposite direction Strict inequality
can occur hold) in (8) only if..... The condition (8) holds with strict inequality.
infer [see also: conclude, deduce, follow] From (5) we infer that..... We have shown that.....,
whence it is readily inferred that..... It can be inferred from known results that these series at
best converge conditionally in L
p
.
innite Leray and Schauder laid the foundations for the generalization of the Lefschetz index
to innite dimensions. Thus F vanishes to innite order at x. Note that both sides of the
inequality may well be innite. The sum in (2), though formally innite, is therefore actually
nite. We consider every subset of N, whether nite or innite, to be an increasing sequence.
innitely The known results show that the situation is innitely more complicated there than
in L
2
. We oer numerical evidence to support a conjecture that there exist innitely many
primes of this type. innitely often an innitely dierentiable function
innity Here the interesting questions are not about individual examples, but about the asymp-
totic behaviour of the set of examples as one or another of the invariants (such as the genus)
goes to innity.
inuence [see also: aect] Inuenced by (2) of Theorem 3, Danes (1989) suggested that.....
MATHEMATICAL ENGLISH USAGE 73
informally Informally said, gaps go to gaps. Stated informally, continuous functions of mea-
surable functions are measurable.
information [see also: data, material] Theorem 1 gives information on about)..... Also, wher-
ever possible, we work with integer coecients, enabling us to obtain information about torsion.
The main information conveyed by this formula is that..... For background information,
see [5]. This accords with the intuition that as we pass down the coding tree, we nd out
more and more detailed information about the ordering actually represented. The interested
reader is referred to [4] for further information. [Note the double r in referred.] Another proof
(yielding more information) can be found in [GH]. The survey article [5] by Diestel contains a
wealth of information about the Dunford-Pettis property. Intuitively, entropy of a partition is
a measure of its information contentthe larger the entropy, the larger the information content.
These three results lead to several illuminating pieces of information about the (insuciently
studied) Berger property in general spaces. [Note that information has no plural and does not
appear with an.]
informative a highly informative book
ingredient [see also: element, feature, detail, characteristic, point, aspect] The main crucial/key)
ingredient in the proof of..... is..... The method sketched in Section 3 of [Con] carries through
with our choice of =
1
+
2
, but there is one extra ingredient worthy of mention.
inherent [see also: essential, basic, intrinsic] It is important to notice some of the weaknesses
inherent in the above approach. We now indicate some of the inherent diculties.
initially [= at rst] The function f (initially dened on C
0
) determines a functional on S.
initiate This research was initiated when the rst author was visiting the University of Alberta
in the summer of 2008; the nancial support and kind hospitality are gratefully acknowledged.
innermost the innermost sum in (4)
inordinately We also need the following technical lemma, which asserts the rarity of numbers
with an inordinately large number of prime factors.
insert [see also: plug, put, place] Inserting additional edges destroys no edges that were already
present.
inside [see also: within] Inside U, the zeros are also quantitatively restricted. The intervals we
are concerned with are either completely inside A or completely inside B. the inside diameter
inspection [see also: check, examination, scrutiny, verication] The theorem is false without
the assumption that....., as an inspection of Example 3 shows. A close inspection of the proof
reveals that..... a careful detailed/thorough/brief/routine) inspection
inspiration [see also: motivation, impulse] The inspiration for Theorem 1 was the paper of
Snow.
inspire [see also: motivate, impulse] This inspired us to take a fresh look at all the results in
[BG].
instead [see also: place, replace, rather] Instead of using the Fourier method we can multiply.....
[Not: Instead using] Instead of dealing with lines one by one, we deal with collections of
lines simultaneously. The proof we give is based instead on Galois theory. Strictly speaking,
74 JERZY TRZECIAK
we should write something like a(l, m, n) to reect the dependence; we shall rely upon context
instead. Rather than working directly with V (s), we shall instead consider the following two
general integrals:..... [Or: Rather than work]
institution We would like to express our thanks to all of these institutions for their hospitality
and support.
integer Then F has simple zeros with residue 1 at the integers. a polynomial with integer
coecients
integrable a p-integrable function the space of pth power integrable functions
integral 1 the integral of f on A over A)
2 the integral multiples of 2
n
integrate We now integrate around the circle over [x[ < 1 /on [x[ < 1) to obtain.....
integration We see with the aid of an integration by parts that..... Theorem 2 makes it
legitimate to apply integration by parts. By k-fold integration by parts,..... This bound is
integrable and on integration one again obtains the same form of..... Functions which are equal
almost everywhere are indistinguishable as far as integration is concerned. Without losing any
generality, we could have restricted our denition of integration to integrals over all of X. [Not:
Without loosing] Recently proofs have been constructed which make no appeal to integration.
To justify the interchange of summation and integration, consider..... The interchange in the
order of integration was legitimate, since..... One is tempted to reverse the order of integrations
but that is illegitimate here.
intend [see also: aim, design] We intend to analyze ideals of the form x
i
) for i N. Condition
(c) is intended to give us rm control over..... Thus the paper is intended to be accessible both
to logicians and to topologists. This volume was originally intended as a celebration volume
marking the occasion of N. Wieners seventieth birthday.
intention [see also: aim, purpose, objective, design] It has not been our intention to claim the
originality of this result.
intentionally [see also: deliberately] This terminology is intentionally provocative.
interchange [see also: exchange] 1 To justify the interchange of summation and integration,
consider..... The interchange in the order of integration was legitimate, since.....
2 It follows, by interchanging the roles of X and Y , that..... The lower limit is dened analo-
gously: simply interchange sup and inf in (1).
interchangeably We shall write H(x, t) and H
t
(x) interchangeably.
interest 1 [see also: importance] The interest of the lemma is in the assertion that..... There
has recently been increasing interest in the theory of..... This article features results in both
spectral theory and operator ergodic theory made possible by a recent renewal of interest in the
consequences of Jamess inequalities. Our interest in extending Strangs results comes from
the fact that..... The authors interest in this problem was recently rekindled by a conversation
with David Lees. The theorem gains in interest if we realize that..... The usefulness and
interest of this correspondence will of course be enhanced if there is a way of returning from the
transforms to the functions, that is to say, if there is an inversion formula. This class is wide
enough to include a number of examples of interest. It is therefore of interest to look at the
MATHEMATICAL ENGLISH USAGE 75
asymptotic behaviour of..... Of particular interest to us is S(5, 3). Canonical products are of
great interest in the study of entire functions of nite order. The polynomials R
i
do not have
this stability property, and are therefore of little interest. We now prove a result that may be of
independent interest. The case n = 2 is of no interest since..... We have made modications
in the interest of readability.
2 We shall be interested in seeing whether..... We now prove a lemma which is interesting in
its own right. The preceding proof contains a result which is interesting enough to be stated
separately. Interestingly, it turns out that there are examples of..... Galois correspondences
are uninteresting from the dynamical point of view.
interplay The interplay of these two kinds of sets will be crucial in the proof of our main theorem.
interpret [not: interprete] Now (1) can be interpreted to mean that A = B. Theorem 3
may be interpreted as saying that A = B, but it must then be remembered that..... Neverthe-
less, in interpreting this conclusion, caution must be exercised because the number of potential
exceptions is huge. The right-hand members of (1) and (2) are to be interpreted by continuity
when f is in Z.
interpretation This inequality admits of a simple interpretation. The interpretation of is
that for every x,..... Of course, a literal interpretation of (1) makes no sense.
interrupt In this section we gather some miscellaneous results that are more or less standard.
These will be used to calculate the constants c

(, ), and we place them here to avoid inter-

rupting the forthcoming arguments.
intersect [see also: meet] The two sets intersect in at most three points. Then one Y
i
can
intersect another only in one point.
intersection Since M is an intersection of closed subspaces, it is a closed subspace of E. Let
M be the intersection of the sets M
i
.
interval The zeros appear at intervals of 2m.
intimately [see also: closely] The above-mentioned measure is of course intimately related to
the geometry of the real line.
into We must now bring dependence on d into the arguments of [9]. The proof will be divided
into a sequence of lemmas. We can factor g into a product of irreducible elements. Other
types t into this pattern as well. This norm makes X into a Banach space. We regard (1)
as a mapping of S
2
into S
2
, with the obvious conventions concerning the point . We can
partition [0, 1] into n intervals by taking..... The map F can be put brought) into this form
by setting..... The problem one runs into, however, is that f need not be..... But if we argue
as in (5), we run into the integral....., which is meaningless as it stands. Thus N separates M
into two disjoint parts. Now (1) splits into the pair of equations..... Substitute this value of
z into in) (7) to obtain..... Replacement of z by 1/z transforms (4) into (5). This can be
translated into the language of dierential forms. Implementation is the task of turning an
algorithm into a computer program.
intricate [= very complicated or detailed; see also: elaborate, involved, sophisticated] It turned
out to have an extremely intricate structure.
intrinsic Our method has the disadvantage of not being intrinsic.
76 JERZY TRZECIAK
introduce [see also: set up] We nd ourselves forced to introduce an extra assumption. In order
to state these conditions succinctly, we introduce the following terminology. We now introduce
the algebras we shall be concerned with. This motivated the second author to introduce the
notion of..... More specialized notions from Banach space theory will be introduced as needed.
introduction The projection technique requires the introduction of an appropriate homomor-
phism. This suggests the introduction of the dierential operator A =..... We now come to
the theorem which was alluded to in the introduction of the present chapter.
intuition It is useful to consider some rather simple examples to gain some intuition. This
accords with the intuition that as we pass down the coding tree, we nd out more and more
detailed information about the ordering actually represented. Our main nding in this paper
is that this intuition turns out to be erroneous.
intuitively Intuitively, entropy of a partition is a measure of its information contentthe larger
the entropy, the larger the information content.
invalid [see also: fallacious, erroneous, fail] This argument is invalid for several reasons.
invariant Clearly, F leaves the subspace M invariant. Thus F is G-invariant invariant under
the action of G). Here the interesting questions are not about individual examples, but about
the asymptotic behaviour of the set of examples as one or another of the invariants (such as the
genus) goes to innity.
inverse [see also: converse, opposite, reverse] Then F is the homeomorphism X Y inverse
to G. No x has more than one inverse. We use upper case letters to represent inverses of
generators. the inverse image = the preimage
investigate [see also: examine, explore, study, test] In this section we investigate under what
conditions the converse holds. The situations with domains other than sectors remain to be
investigated.
investigation [see also: study, research, analysis, exploration] We shall pursue our investigation
of conservation laws in Section 5. This paper, for the most part, continues this line of inves-
tigation. The results have been encouraging enough to merit further investigation. a careful
close/detailed/extensive) investigation
invoke [see also: appeal, call upon] Finally, case (E) is completed by again invoking Theorem 1.
This allows us to invoke Theorem B to produce an ultralter satisfying.....
involve [see also: entail, necessitate, require] Our proof involves looking at..... This involves no
loss of generality. The analysis of PDO involves the geometry of T

M. The terms involving

a ,= 1 contribute O(n), since..... Perhaps the most important problem involving f-vectors is
whether or not Johns conditions extend to spheres. A computational restraint is the algebraic
number theory involved in nding these ranks, which will typically be more demanding than in
our example of Section 1. The crucial set-theoretic properties involved are the following. [Note
the position of involved; placed before properties, it would mean complicated.]
involved [see also: dicult, hard, complex, complicated, intricate, sophisticated] It seems likely
that the arguments would be much more involved. Besides being very involved, the proof gives
no information on.....
MATHEMATICAL ENGLISH USAGE 77
inward(s) the inward pointing unit normal to M [In most adverbial uses, inward and inwards
are interchangeable; as an adjective, the more standard form is inward: the inward layer.]
irrelevant [to sth] We shall see later that the values of h(n) for large n are irrelevant to the
problem. The choice of A is clearly irrelevant, so assume A = 0.
irrespective [of sth; see also: regardless] The method works irrespective of whether A or B is
used.
issue [see also: matter, subject, topic, point] One delicate issue is that it is not known whether.....
To get around this delicate issue, we shall separate the variables s and u. Addressing this
issue requires using the convergence properties of Fourier series. However, we prefer to avoid
this issue altogether by neglecting the contribution of B to S. Since RG-modules in general
need not be completely reducible, indecomposability becomes an important issue. a central
critical/crucial/key/main/major/wider/minor/side/basic/fundamental/complex) issue
it [see also: this] It is Proposition 8 that makes this denition allowable. It is this point of
view which is close to that used in C

-algebras. It is a theorem of Watanabe that A is an ideal

in A

if and only if..... It is these buers that make it easy to see that the limit continuum
is contractible. Of course, it is tacitly understood that it is this measure that is really under
discussion. It has to be assumed that..... Theorem 2 makes it legitimate to apply integration
by parts. We leave it to the reader to verify that..... We shall nd it convenient not to
distinguish between two such sequences which dier only by a string of zeros at the end.
But....., it being impossible to make A and B intersect. [= since it is impossible to make]
item [see also: point, matter, issue] Since most of the results presented are quite classical (the
novelty lies in the arrangement, and some of the proofs are new), I have not attempted to
document the source of every item. The third item is proved similarly.
iterate 1 One can iterate Theorem 2 to conclude that.....
2 The iterates eventually reach the value 1.
itself This itself does not produce a solution of (1), but an additional hypothesis such as the
Palais-Smale condition does provide such a solution. We have to show that M itself is an
algebra. [Or: M is itself an algebra.] But H itself can equally well be a member of S. Now
f is independent of the choice of (although the integral itself is not). Thus in G itself there
can be no such equivalences with f(r) = 2.
J
job Two more triangulations will nish the job.
join [see also: connect, relate, tie, associate] We can join a to b by a path . [Or: join a and b;
not: join a with b] Parallel lines never join. These subsets join up to form a simple closed
curve passing through A and B.
just [see also: equally, merely, only, simply, exactly, precisely] His techniques work just as well
for general v. The other inequality is just as easy to prove. When n = 0, (7) just amounts to
saying that..... We shall be brief here and just estimate the terms that are the most challenging.
When there is no ambiguity we drop the dependence on B and write just Y
T
for Y
T,B
. Now,
just the fact that F is a homeomorphism lets us prove that..... Can f(x) > 1 be replaced by
just x > 1? The inverse image of A consists of just the basepoint of X. In fact, we shall prove
78 JERZY TRZECIAK
our result under the weaker hypothesis that W is weakly bounded, rather than just bounded, on
an innite subset of G. This follows from Lemma 2 just the way (a) follows from (b). This
class includes just under 3000 items.
justify [see also: legitimate, meaningful, explain, reason] To justify the interchange of summa-
tion and integration, consider..... The last but one step is justied by the fact that.....
juxtaposition Concatenation of sequences is indicated by juxtaposition in the obvious way.
K
keep [see also: leave, retain, preserve, maintain, stick] Keep only those vertices whose coordi-
nates sum to 4. The mapping f keeps the origin xed. The proof is mainly included to keep
the exposition as self-contained as possible. Keep in mind that we are now using algebraic
notation. We have to keep track of how the constant K depends on the domain D.
key [see also: basic, critical, crucial, essential, main] The key observation is that if..... The key
part is to show that the submanifolds U
k
t together to form a complex submanifold. A key
step in obtaining (6) is Jensens inequality. It supplies the key to the proof of Theorem 2.
The following lemma is the key to extending Wagners results.
kind [see also: sort, type, form] Another kind of modication is illustrated in the next lemma.
The same kind of approach has been taken successfully to determine all regular maps of various
small genera. processes of an entirely new kind a connection of some kind
know [see also: realize, recognize, grasp, aware] We do not know whether or not Q(R) = R in
this situation. The map f, which we know to be bounded, is also right-continuous. Knowing
this matrix is equivalent to knowing the multiplicities of the k
i
. However, we know of no way
of deriving one theory directly from the other. All the Cox maps F are known to have L(F)
nite. The best known of these is the Knaster continuum. The answer is not known to us.
The only references known to the authors are [A] and [V], where the case A = L(E) is settled in
the negative. Many of them were already known to Gauss. It has long been known that.....
It has been known for some time that..... The conjecture (now known not to be true in
general) was that..... So far it seems not to be known whether the geometric condition on X
can be omitted. In particular, there is a summary of what is known about polarized pairs
with small genus. The importance of these examples lay not only in lowering the dimension of
known counterexamples, but also in..... [Note that the past tense of lie is lay, not lied.] The
following proposition is probably well known, but we do not have a reference.
knowledge To the best of the authors knowledge, the problem is still open. However, to our
knowledge this is not fully resolved. Indeed, to our knowledge, cardinality restrictions on Berg
spaces have received no prior attention outside the metric context. Give a proof of Theorem
2 which requires no knowledge of the boundary values of f. The knowledge of the invariant
subspaces of an operator helps us to visualize its action. We presume a basic knowledge of
large cardinals and forcing.
L
label 1 The edges F and G have opposite labels. The vertex F has label 1.
2 [see also: indicate, mark] The point F is labelled 1. [Or: labelled by 1.] We may assume
after passing to a subsequence (still labelled v
n
) that.....
MATHEMATICAL ENGLISH USAGE 79
lack 1 [see also: absence] Note that the lack of compactness prevents us from using Theorem 1.
Thus, although we follow the general pattern of proof of Theorem A, we must also introduce new
ideas to deal with the lack of product structure. [Note the dierence: absence = non-presence;
lack = shortage of something desirable.]
2 The space R is pseudo-compact but lacks any such subset.
language [see also: term] This can be translated into the language of dierential forms.
Rephrased in the language of [HT], Proposition 2 says that..... In geometric language, the
hypothesis is that F is.....; part of the conclusion is that F satises.....
large [see also: great, more, extensive, substantial] This is a condition on how large f is. The
process slows down for large x. Unfortunately, this only holds for n suciently large. Then
M
n
(x) = 0 for every x, however large. [= no matter how large] Here L can be taken as large
as desired. There are a large number of examples showing that..... [Not: There is a large
number] The kernel satises good large time bounds if..... But A
n
z
n
is much larger than
the sum of all the other terms in the series

A
k
z
k
. The question of whether B is ever strictly
larger than A remains open. Intuitively, entropy of a partition is a measure of its information
contentthe larger the entropy, the larger the information content. the second largest element
By and large, we shall use the same notation as in..... [= In general]
largely [see also: mostly, mainly, essentially, broadly] This eect is largely [= mostly] due to
the presence of the logarithmic factor. The remainder of the estimation is largely [= to a great
extent] as before with B replaced by C. Over the past ten years the isomorphic structure of
spaces of weighted holomorphic functions has been largely [= almost completely] determined.
We will nish this section by oering a second application of our machinery (although in truth
it is largely a corollary of the above).
last The vector v has at least n ones in its last m positions. This last result means that.....
This last means that..... These last two are called the Banach maps. (see the last paragraph
but one of page 24) The last but one step is justied by the fact that..... the next-to-last
inequality = the second last inequality = the penultimate inequality the third last row = the
antepenultimate row the last-mentioned map Now (d) is clear from the very last statement
of Theorem 4. He accounts for all the major achievements in topology over the last few years.
We can at last show that the limit in (1) exists. [Note that at last means after much delay;
to introduce a last point or reason, use nally or lastly.]
lastly [see also: nally, last] And lastly, what do we mean by an admissible map?
late [see also: subsequently] For later use, we prove a mild renement of this latter characteri-
zation. Let us note (for later reference) that..... Later, by bringing in the injectivity radius,
Fong simplied the argument. We shall see later that the values of h(n) for large n are irrele-
vant to the problem. We now state a result that will be of use later. To save space later on,
all modules are given in the form.....
latter [,= later; see also: former] Both F and G are connected, but the latter is in addition
compact. Then either..... or..... In the latter case,..... Examples 1 and 2 give two operators,
the former bounded and the latter not, with.....
lay The conference laid the basis for a series of annual gatherings. In his Stony Brook lectures,
he laid great emphasis on the use of..... In Section 2, we lay the foundations for a systematic
80 JERZY TRZECIAK
study of..... Leray and Schauder laid the foundations for the generalization of the Lefschetz
index to innite dimensions. There are some obvious necessary conditions, rst laid in [8].
[= presented] We lay out the details of this generalization in the rst part of this paper.
lead This leads to the contradiction that 0 < a < b = 0. Interchanging the roles of f and g
leads from (5) to..... These properties led him to suggest that..... This leads us to consider the
closed subscheme dened by G. The above construction has led the author to believe that.....
So one is naturally led to an investigation of..... If we simply mimic the standard proof of.....,
we are led to.....
learn [see also: nd out, discover] In the course of writing this paper we learned that P. Fox
has simultaneously obtained results similar to ours in certain respects. However, shortly after
learning about Wieners work, P. Levy found a more elementary argument.
least [see also: minimal] This is the least useful of the four theorems. The least such constant
is called the norm of G. The rank of appearance (or simply the rank) of an integer in the
sequence U(P) is the least positive integer n such that..... We select a system for which the
power of b dividing n is least. That is the least one can expect. It does not matter in the
least whether..... Each f lies in zA for at least one A. The degree of F exceeds that of G by
at least 4.
leave [see also: keep, remain, preserve] Then we can nd some net (s
k
) which eventually leaves
every compact subset of G. We leave the details to the reader. We leave it to the reader to
verify that..... The mapping f leaves the origin xed. These results leave open the basic case
k = . So s can be thought of as q with F
q
extended but X
q
left the same. What is left is
to show that..... We are left with the task of determining.....
left [see also: right] Examination of the left and right members of (1) shows that..... Under
those conditions, what does the sum on the left hand side of (8) signify? Note that (2) serves
as the denition of its left side. We denote the left-hand such interval by I
L
and the right-hand
one by I
R
. Now F is dened to make G and H match up at the left end of I. The tangent
space to N at x is identied with M via left translation. The method of proof of Theorem B
can be adapted to extend the right-to-left direction of Mostowskis result by showing that.....
Comparisons are done in left-to-right order. The entries in each row are increasing from left
to right. The string N (read from right to left) starts with..... We say that F is low if F(c)
lies to the left of c. To be precise, A is only left-continuous at 0.
legitimate [see also: justify, meaningful] Theorem 2 makes it legitimate to apply integration by
parts. The denition is legitimate, because..... The interchange in the order of integration was
legitimate, since..... In the preceding proof, the appeal to the dominated convergence theorem
may seem to be illegitimate, since.....
lemma [pl. lemmas or lemmata] We shall prove this theorem shortly, but rst we need a key
lemma. The proof will be divided into a sequence of lemmas. We now prove a lemma which
is interesting in its own right. The interest of the lemma is in the assertion that..... We defer
the proof of the moreover statement in Theorem 5 until after the proof of the lemma. With
Lemma 4 in at) hand, we can nally dene E to be equal to P(m)/H. ....., from which it
is an easy step, via Lemma 1, to the conclusion that..... The following lemma is the key to
extending Wagners results. The following lemma, crucial to Theorem 2, is also implicit in [4].
Note that this lemma does not give a simple criterion for deciding whether a given topology is
MATHEMATICAL ENGLISH USAGE 81
indeed of the form T
f
. At rst glance Lemma 2 seems to yield four possible outcomes. The
nal lemma is due to F. Black and is included with his kind permission.
lend One of the appealing aspects of the spectral set is that it readily lends itself to explicit
computation. This lends precision to an old assertion of Dini:.....
length [see also: detail, depth, expand] Pick the rst arc of length 1 in this sequence. The
interval J has length 2k. This subject is treated at length in Section 2. We shall discuss this
again at somewhat greater length in Section 2.1.
lengthy A rather lengthy calculation shows that.....
less [see also: few] Then F is 2 less than G. Let A
n
be a sequence of positive integers none of
which is 1 less than a power of two. Thus F is less than or equal to G. [Not: less or equal to
G, nor less than or equal G] Here F is strictly less than G. Thus F is no less than G.
Clearly, F is less than 1 in absolute value. Less than 1 in p of its points will result in a quartic
with ideal class number p. [Do not write: X has no less elements than Y has if you mean:
X has no fewer elements than Y has; less should not be followed by a plural countable noun.
However, use less when it is followed by than or when it appears after a noun: X has no less
than twenty elements; Y has ten elements or less.] Much less is known about hyperbolically
convex functions. Although our proof is a little tedious, it is much less so than Itos original
proof, which was carried out without the benet of martingale technology. This method is
recently less and less used. to in) a lesser degree
let Let R be a ring and A a right R-module. [Not: Let R a ring] Let f satisfy (2). [Not:
Let f satises (2), nor Let f verify (2).] Let f be the linear form g (m, g). We let T
denote the set of..... One cannot in general let A be an arbitrary substructure here. Letting
m tend to zero identies this limit as H. As we let t vary, f(t) describes a curve in M. The
desired conclusion follows after one divides by t and lets t tend to 0. Now, just the fact that
F is a homeomorphism lets us prove that.....
letter We use the letter m for..... The letter will be reserved for characteristic functions
throughout this book.
level [see also: degree, stage] Two words are m-equivalent if the corresponding paths terminate
at level m at the same vertex and coincide from level m+ 1 to n. These results show that an
analysis purely at the level of functions cannot be useful for describing..... The vector eld H
always points towards the higher A-level. The temperature has to be maintained at a very high
level. Ination in the rst quarter rose beyond the acceptable level of 5%. The level of.....
has remained largely unchanged for many years. Prewar levels of production were surpassed
in 1929.
lie Each component which meets X lies entirely within Y . Then C lies on no segment both of
whose endpoints lie in K. The advantage of using..... lies in the fact that..... This simple
observation lies at the heart of the proof of our rst theorem. The importance of these examples
lay not only in lowering the dimension of known counterexamples, but also in..... [Note that the
past tense of lie is lay, not lied.]
light These facts shed new light a new light) on..... From the standpoint of linear programming,
the above discussion is incomplete in that it throws no light upon the question whether the
82 JERZY TRZECIAK
function F attains its inmum. In light of In the light of) the obvious inequality a b,.....
This is no coincidence, in the light of the remarks preceding Denition 2.
like [see also: as, resemble, similar] Thus modules over categories are in many ways like ordinary
modules. So we must in particular show that sets like this are not added. It should come as
no surprise that a condition like a
i
,= b
i
turns up in this theorem. A formula like (3) surely
deserves some explanation. Specically, one might hope that a clever application of something
like Choquets theorem would yield the desired conclusion. Construct an example, like that of
Example 9, in which (1) fails but (2) holds. It is now apparent what the solution for K will be
like:..... Let us see what such a formula might look like, by analogy with Fourier series. The
proof is nonconstructive and gives no indication of what the exceptional set may look like.
likely [see also: possible, plausible] All inputs of size n are equally likely to occur. It seems
likely that the arguments would be much more involved. It is highly likely that if one of the
Xs is exchanged for another, the inequality fails. A complete explication of the Fox spaces
is warranted, as it will likely reveal further clues to the dierences between the parabolic and
hyperbolic theories. This change is unlikely to aect the solution. It is unlikely that the
disturbances will eventually disappear.
likewise [see also: similarly, also, moreover] Likewise, if A does not span C(I), removal of any
of its elements will diminish the span.
limit 1 The limit lim
x0
f(x) exists. [Not: There exists a limit lim
x0
f(x).] Now (1) follows
after passage to the limit as n .
2 [see also: conne, restrict] I shall limit myself to three aspects of the subject.
limitation [see also: restriction] All these methods had severe limitations.
line [see also: approach, procedure] The proof proceeds along the same lines as the proof of
Theorem 5, but the details are more complicated. Following the same lines we nd that it takes
k prolongations to get an immersed curve. For direct constructions along more classical lines,
see [2]. Although these proofs run along similar lines, there are subtle adjustments necessary
to t the argument to each new situation. Further results along these lines were obtained by
Clark [4]. This paper, for the most part, continues this line of investigation. The same line
of reasoning applies in the continuous time setting. Here c denotes a constant which can vary
from line to line. Then....., where the fact that A = B was used on the penultimate line. The
inmum in the nal line here is equal to S. a broken dashed, dotted, slanting, wavy) line
link 1 [see also: connection, relationship] There is a close direct/strong) link between..... We
also make links with some classical notions, in particular, Taylors structure semigroup. It is
worth making a link with Theorem 1.
2 [with/to sth; see also: connect, join, relate, tie, associate] The spectral theory of..... is closely
linked to the discrete Hilbert transform.
list 1 Let f be the ith formula on the list. The algorithm compares x with each entry in turn
until a match is found or the list is exhausted.
2 [see also: enumerate] In the nal section of the paper, we list some open problems. All
possible types are listed in Table 4.
literal Of course, a literal interpretation of (1) makes no sense.
MATHEMATICAL ENGLISH USAGE 83
literature There is now an extensive literature dealing with..... For a comprehensive treatment
and for references to the extensive literature on the subject one may refer to the book [M] by
Markov. However, no extension in this direction has appeared in the literature. Inciden-
tally, the question of whether K(E) is amenable for specic Banach spaces E seems to have
received almost no attention in the literature. Unfortunately, the connection with Gromovs
work has been obscured in recent years by an emphasis (in the algebraic topology literature) on
conguration spaces.
little [see also: minor, slight, somewhat, hardly] The polynomials R
i
do not have this stability
property, and are therefore of little interest. This idea is very little dierent from what can
already be found in [2]. Apart from a few embellishments necessitated by some technical
diculties, the ideas dier very little from those used to prove Lemma 4. With a little more
work we can prove..... Little else is known about the Klein property in this class of spaces.
locate Let us digress to locate the need for relative ergodicity in the above arguments. These
slits are located on circles about the origin of radii r
k
.
location [see also: place, position] Theorem 7 imposes a quantitative restriction on the location
of the zeros of..... The location of the zeros of a holomorphic function in a region is subject
to no restriction except the obvious one concerning the absence of limit points in . Let n
k
be
the rst location to the right of the kth decimal place of W that has a value less than b.
long Hence F is twice as long as G. However, with the recent advent of simulation based
inference, the need for analytically tractable posteriors is no longer critical. It turns out that
this is independent of the representations taken (as long as they are faithful). [= provided that
they are faithful] It has long been known that..... The following simple example has long
been a part of ergodic-theoretic folklore. a long-standing conjecture
look 1 [see also: glance, inspection, scrutiny] Let us now take a quick look at the class N, with
the purpose of determining how much of Theorems 1 and 2 is true here. This inspired us to
take a fresh look at all the results in [BG]. A careful look at the proofs reveals that.....
2 [see also: view, regard, examine, study, appear, seem] The preceding observation, when
looked at from a more general point of view, leads to..... Rather than discuss this in full
generality, let us look at a particular situation of this kind. Suppose, to look at a more specic
situation, that..... The next lemma shows how such a semilattice looks when embedded in a
larger compact semilattice. Let us see what such a formula might look like, by analogy with
Fourier series. The proof is nonconstructive and gives no indication of what the exceptional set
may look like. This observation prompted the author to look for a more constructive solution.
Here, of course, the set A produced is rather thin and certainly nowhere near the densities we
are looking for.
lose This extension retains control on..... at the sacrice of losing some control on..... [Not:
loosing] The total amount of information lost is.....
loss [see also: harm] It simplies the argument, and causes no loss of generality, to assume.....
This involves no loss of generality. Without loss of generality we can assume that..... [In
many cases, the phrase without loss of generality can be omitted; write simply: We can assume
that..... Avoid using the abbreviation w.l.o.g.]
lot [see also: heavily] This is an area where there is currently a lot of activity.
84 JERZY TRZECIAK
low [see also: lower] The weight dropped to a low of 7 kg.
lower 1 [see also: decrease, diminish, reduce, cut down, limit] The importance of these examples
lay not only in lowering the dimension of known counterexamples, but also in..... [Note that the
past tense of lie is lay, not lied.]
2 The lower limit is dened analogously: simply interchange sup and inf in (1). The reader
may wonder why we have apparently ignored the possibility of obtaining a better lower bound
by considering..... a path obtained by going from A to B along the lower half of the circle
a lower semicontinuous function
M
machinery We will nish this section by oering a second application of our machinery (although
in truth it is largely a corollary of the above).
magnitude [see also: size] In all our analysis, only the order of magnitude of P will be signicant.
main [see also: critical, crucial, essential, key, major, primary, principal] The main problems
that we address are..... Our main results state in short that MEP characterizes type 2 spaces
among reexive Banach spaces. In brief outline, here is the main idea of the proof. The main
new feature is the use of the face ring to produce lower bounds for the number of vertices.
mainly [see also: mostly, largely, primarily, part] The proof is mainly included to keep the
exposition as self-contained as possible. Although [1] deals mainly with the unit disc, most
proofs are so constructed that they apply to more general situations. We conclude with two
simple lemmas to be used mainly in.....
maintain [see also: keep, retain, preserve] We maintain the convention that implied constants
depend only on n. The temperature has to be maintained at a very high level. Now it is
a simple matter to change the denition of the F
i
at the single point zero, still maintaining
condition (C), so that F is no longer discontinuous.
major [see also: important, principal, main, signicant] One major advantage of..... is that.....
This was one of the major steps in Wieners original proof of his Tauberian theorem.
majority [see also: most] There is, however, a simple condition, satised in the vast majority
of applications, which ensures.....
make [see also: do, build, construct, perform, carry out, produce, convert, transform, turn into,
force, compose, comprise, consist, constitute] The tensor product makes G a module over R.
In [2], this theorem is made the starting point of Gelfand theory. This norm makes X into
a Banach space. Theorem 2 makes it legitimate to apply integration by parts. Now (8)
makes it obvious that..... This device makes it possible to replace multivalued functions by
functions with..... .....where C can be made arbitrarily small by taking..... If this is not so,
a linear fractional transformation will make it so. We make G act trivially on Y . Now F
is dened to make G and H match up at the left end of I. But..... it being impossible to
make A and B intersect. [= since it is impossible to make.....] The denition of generator
is designed to make the proof above work for M = Z. The function of Lemma 2 can be
made to satisfy..... A similar reformulation can be made for..... It is sucient to make the
computation for T. After making a linear transformation, (9) becomes..... Let us make the
following observation assumption/denition). We make the convention that f(Q) = i(Q).
MATHEMATICAL ENGLISH USAGE 85
Recently proofs have been constructed which make no appeal to integration. It is worth
making a link with Theorem 1. Nevertheless, it might be possible to make sense of (2) even
for non-injective V by considering a multi-valued operator Z. Indeed, it is routine to verify
that the index so constructed is independent of the choices made. The set WF(u) is made up
of bicharacteristic strips. Each of the terms that make up G(t) is well dened. Women make
up two-fths of the labour force.
manage Suppose we have managed to nd A
i
for all i < n.
manifestly [= evidently] This series is manifestly convergent.
manipulate [see also: manage, control, handle] While nonparametric priors are typically dicult
to manipulate, we believe the contrary is true for quantile pyramids.
manner [see also: fashion, way, method] Theorem 2, at the end of Section 2, was not originally
obtained in the manner indicated there. If our measure happens to be complete, we can dene
f on E in a perfectly arbitrary manner. It is an easy matter to use Theorem 10 to construct
all manner of interesting Peano continua [= continua of dierent kinds]
many [see also: abound, number, numerous, profusion, several, variety, abundance] Many of
them were already known to Gauss. The proof makes use of many of the ideas of the general
case, but in a simpler setting. Thus G has 10 normal subgroups and as many non-normal ones.
Consequently, H is a free R-module on as many generators as there are path components
of X. Therefore, A has two elements too many. [Or: A has two too many elements.] Then
A has three times as many elements as B has. It meets only countably many of the Y
i
.
a sequence with only nitely many terms nonzero To compute how many such solutions there
are, observe..... How many of them are convex? How many such expressions are there?
How many entries are there in this section? How many multiplications are done on average?
How many zeros can f have in the disc D?
map 1 an onto map a one-to-one map The continuum Y is tree-like since it admits a map
onto X. This map carries lines to lines carries M onto N).
2 [see also: send] F maps a, b to c, d. The point x maps to under f.
mark 1 [see also: sign] The mark over an element denotes that it has been omitted.
2 [see also: indicate, label] The arrow marks the direction of the resulting ow. The greatest
element is marked with a dot. In Figure 2, the set A is marked by a square with a small
triangle on top. Number the successive segments of the boundary line between A and B
(marked thickly in the picture) with the numbers 0, 1, ..., n, starting at the bottom. a marked
dierence [= obvious, noticeable]
match 1 The algorithm compares x with each entry in turn until a match is found or the list is
exhausted.
2 [see also: accord, agree, correspond] Suppose the rst three characters of the pattern match
three consecutive text characters. Now F is dened to make G and H match up at the left
end of I.
material [see also: information, data, fact, detail] Basic material on semigroups of operators
can be found in [4]. For relevant background material concerning random walks, see [2].
Section 2 contains some specic preparatory material, notably a brief discussion of the category
of virtual vector bundles.
86 JERZY TRZECIAK
matrix Dene A to be the matrix with 1 in the (i, j) entry and 0 elsewhere. Take A to be the
matrix with all entries zero except for i j at (i, j). a real n n matrix
matter 1 [see also: issue, subject, topic, thing, problem, point] This proves one half of (2); the
other half is a matter of direct computation. The rst of the above equalities is a matter of
denition, and the second follows from (3). It is a simple matter a routine matter/a matter
of routine) to show that..... It is an easy matter to use Theorem 10 to construct all manner
of interesting Peano continua [= continua of dierent kinds] We conclude that, no matter
what the class of b is, we have an upper bound on M. Nevertheless, no matter how small a
neighbourhood of x we take, the image will be..... However, for three more subgroups, matters
become more complicated.
2 [see also: important, relevant] It does not matter in the least whether..... Whether or not
this is correct does not matter; we are trying to motivate the proof that follows.
maximal [see also: maximum] Suppose A is maximal with respect to having connected preim-
age. Take N to be a family of normal measures in P(X) such that N is maximal subject to the
condition that the supports of the measures in the family are pairwise disjoint. Now suppose
that F(n) = x and that n is maximal in this respect.
maximum [see also: maximal, supremum] The function g attains takes/achieves) its maximum
at x = 5. For example, F reaches a relative maximum of 5.2 at about x = 2.1. Now (c)
asserts only that the overall maximum of f on U is attained at some point of the boundary.
By computing the second derivative we note that x = 1 is a maximum point for f. This has a
maximum value of 4 when x = 2. Then V (x) is the maximum value of J
x
(v) over all controls v.
Let q be the maximum number of variables occurring in..... [Note the double r in occurring.]
vectors in V at maximum distance from v the maximum possible density
may [see also: can, could, might] Then F may or may not x B. Here S may be P, but it may
also not be P. In other algorithms, this may not be true. In addition to a contribution to W
1
,
there may also be one to W
2
. Note that both sides of the inequality may well be innite. It
may well be that no optimal time exists, as the following example shows. It may seem strange
to dene 0. = 0. We may (and do) assume that..... Since the integrands vanish at 0, we
may as well assume that..... However, F is only nonnegative rather than strictly positive, as
one may have expected. Since Z is a nite set, we may continue subtracting suitable scalar
multiples of the x
i
from x. [Or: we can continue]
mean 1 Indeed, N is a Gaussian random variable with mean 0 and variance g.
2 [see also: signify, indicate, convey, suggest] We partially order M by declaring X < Y to
mean that..... Here (1) can be interpreted to mean that..... The rst equality is understood
to mean that..... In Chapter 5, we shall explain what it means for a subset V of A to be
determining for the centre of X. Note that (A) means precisely that condition (B) is not
satised. Important analytic dierences appear when one writes down precisely what is meant
by..... Then (6) merely means that..... The diculty is that it is by no means clear what one
should mean by a normal family.
meaning [see also: sense] We shall also refer to a point as backward nonsingular, with the
obvious analogous meaning. As M is ordered, we have no diculty in assigning a meaning to
(a, b).
MATHEMATICAL ENGLISH USAGE 87
meaningful [see also: justify, legitimate, meaningless] This shows that the sequence (1) is
bounded below, and so the denition of L(f) is meaningful.
meaningless [see also: meaningful] But if we argue as in (5), we run into the integral....., which
is meaningless as it stands.
means [see also: application, device, use, via] This provides an eective means for computing
the index. By relatively straightforward means one can show that..... Then F and G are
homotopic by means of a homotopy H such that..... It is easy to see, by means of an example,
that..... Find integral formulas by means of which the coecients c
n
can be computed from f.
The diculty is that it is by no means clear what one should mean by a normal family.
measure 1 Here dx stands for Lebesgue measure. [Or: the Lebesgue measure] Each set A
carries a product measure.
2 The aim of this article is to study the relationship between the size of A, as measured by its
diameter, and the extent to which A fails to be convex.
meet [see also: intersect, encounter, come across, run into, satisfy] The sets A and B meet in
two points. [= Their intersection is a two-point set] We may assume that this is the rst point
at which these two curves have met. Each component which meets X lies entirely within Y .
The remaining requirements for a type F map are also met. We can also appeal to Lemma 5
to see that the uniform continuity condition (5.3) is met.
member [see also: element, side] Moreover, x is the set whose only member is x. Exami-
nation of the left and right members of (1) shows that.....
membership Computing f(y) can be done by enumerating A(y) and testing each element for
membership in C. Thus W satises two of the four requirements for membership in Z.
mention 1 We conclude with a brief mention of free inverse semigroups. Throughout what
follows, we shall freely use without explicit mention the elementary fact that..... We will work
exclusively in the category of standard Borel probability spaces, and so will often suppress
mention of their sigma-algebras. The method sketched in Section 3 of [Con] carries through
with our choice of =
1
+
2
, but there is one extra ingredient worthy of mention.
2 [sth; doing sth; see also: allude, refer] In [5] he mentions having proved that for g 1. One
more case merits mentioning here. It seems appropriate to mention in passing the corresponding
formula for the cohomology of B
n
. There has since been a series of improvements, of which we
briey mention the work of Levinson. the above-mentioned element the last-mentioned map
the problem mentioned [Not: the mentioned problem]
mere [see also: just, only] The quantity A was greater by a mere 20%. However, this equality
turned out to be a mere coincidence. This result shows that the mere existence of a nontrivial
automorphism j of M produces the cut I(j) of M that satises (2).
merely [see also: just, only] It turns out that A is not merely symmetric, but actually selfadjoint.
But (3) is merely an abbreviation for the statement that..... Here (6) merely means that.....
How is the result aected if we assume merely that f is bounded?
merit 1 [see also: advantage, benet] The merit of Theorem 5 is that it identies.....
2 [see also: worth, warrant] One more case merits mentioning here. The results have been
encouraging enough to merit further investigation.
88 JERZY TRZECIAK
method [see also: approach, device, means, procedure, strategy, technique, way] His method of
proof was to rst exhibit a map..... The method of proof carries over to domains satisfying.....
We require a method of dividing z into subwords z
i
for which we know the structure of
f(z
i
). This provides a method for recognizing pure injective modules. Our method has the
disadvantage of not being intrinsic. All these methods had severe limitations. A shortcoming
of our method is the inability to compare three or more progressions. To show the greater
simplicity of our method over Browns, let us..... In [1] the methods used are those of dierential
topology.
middle We divide them into three classes depending on the width of the middle term as compared
with the width of the other terms. Let E be Cantors familiar middle thirds set.
might [see also: may, could] One might hope that this method would work at least for suciently
regular maps; however,..... It will eventually appear that the results are much more satisfactory
than one might expect. Let us see what such a formula might look like, by analogy with..... .
One must also be aware that the curvature of M
i
might not be bounded uniformly in i.
mild [see also: moderate, reasonable] For later use, we prove a mild renement of this latter
characterization. under mild restrictions on f
mimic If we simply mimic the standard proof of..... we are led to.....
mind [see also: remember] With Lemma 2 in mind, we make the following ad hoc denition.
An example to bear in mind is behaviour in the basin of a periodic point. Two necessary
conditions come to mind immediately. When reading the proof of Lemma 2, it is helpful to
keep in mind that..... It is helpful to keep these similarities in mind.
minimal [see also: least, maximal] In the next theorem, we give fairly minimal conditions that
imply..... Suppose that of all such solutions, (x, y, z) is one with y minimal. Dene n to be
minimal such that f
n
(x) = y.
minimum [see also: maximum] Under what conditions can f have a local minimum in the
interval A at the point x
0
)? This is the minimum property of the Jordan decomposition
which was mentioned in Section 6.
minor [see also: slight, ne] We take this opportunity to correct a minor error in Lemma 2 of
[PS]. with minor modications
minus [see also: plus, remove, delete, negative] It follows that the last sum is equal to minus
itself, and is therefore zero. the minus sign a conformal mapping of Q onto the complex plane
minus the nonnegative real axis
miscellaneous [see also: various] In this section we gather some miscellaneous results that are
more or less standard. These will be used to calculate the constants c

(, ), and we place them

here to avoid interrupting the forthcoming arguments.
miss [see also: avoid, overlook] Half of the sets of R miss i and half the remaining miss j. The
image of U under f misses out more than three points of the sphere.
mistake [see also: error] Note that there is a mistake on p. 3 of [5], where the condition.....
should be deleted.
mnemonic The term upath is a mnemonic for unit speed path.
MATHEMATICAL ENGLISH USAGE 89
model 1 A model for analysing rank data obtained from several observers is proposed. Here we
consider a dual variational formulation which can be derived similarly to that for the sandpile
model. One reason for using this rather elaborate model is that it permits a simple and concrete
denition of the realization. We motivate our work in Lipschitz domains by rst considering a
model case, that of harmonic functions dened on the upper half-space.
2 We model the intensity process for subject i as f
i
=.....
moderate [see also: mild, reasonable] a function of moderate growth
modication [see also: adaptation, adjustment, change, variant, variation, version] The proof
is similar to the proof of Theorem 4, with two principal modications. We write = .....,
which is a slight modication to the previous version of . The only modication to be made
is that we dene..... with minor modications (with the usual modication for p = )
a considerable extensive/major/signicant) modication
modify [see also: alter, change, adjust, vary, revise] The function of Lemma 2 can be easily
modied to obey the extra condition. If h is modied by the addition of a suitable constant, it
follows that..... We nish by mentioning that, suitably modied, the results of Section 2 apply
to the AP case. To obtain the required map one must modify the method as explained in [9].
modulus [pl. moduli] an element of modulus 1 Then F is bounded in modulus by 1. Hence
F is the largest (in modulus) limit point of the sequence F
n
.
moment [see also: time, temporary, temporarily, shortly] A moments consideration will show
that..... Suppose for the moment that q = 1, so that = 1. We shall show in a moment that
these solutions are unique.
monotone [= monotonic; not: monotonous] the monotone convergence theorem [Note that
the term decreasing function is preferable to monotone decreasing function.]
more [see also: additional, extra, further, great] We conclude this section with one more result.
One more case merits mentioning here. Applying this argument k more times, we obtain.....
Then F has n or more zeros in . In dimensions three or more, this would imply..... Beware
that more than one equation is referred to as the fast diusion equation in the literature. If c is
only slightly larger than 1, say 1 < c < 3, then there is more than one curve with this property.
The L
2
theory has more symmetry than is the case in L
1
. a set of no more than k elements
Those more than half a square count as whole ones. ....., which, by another theorem of
Kimneys, is more than enough to guarantee that P gives A outer measure 1. A more complete
theory may be obtained by allowing..... a succession of more and more rened discrete models
This accords with the intuition that as we pass down the coding tree, we nd out more and more
detailed information about the ordering actually represented. How many more men attended
the meeting than the women? The new X is no more continuous, although it still has norm 1.
moreover [see also: also, furthermore, likewise] Moreover, H is a free R-module on as many
generators as there are path components of X. Since A = B and B = C, and since moreover
C = D, it follows that A = C. We defer the proof of the moreover statement in Theorem 5
until after the proof of the lemma.
-morphic As noted before, there exists N homeomorphic to P such that..... This space is
naturally isomorphic to B. Over the past ten years the isomorphic structure of spaces of
weighted holomorphic functions has been largely [= almost completely] determined.
90 JERZY TRZECIAK
-morphically It follows that F maps L homeomorphically onto itself.
-morphism The function thus dened is a semigroup morphism. We turn the set of..... into a
category by dening the morphisms to be..... By a partial automorphism of A we understand an
isomorphism f : B C between two subalgebras B and C of A. The induced homomorphism
is multiplication by 2. If we know a covering space E of X then not only do we know that.....
but we can also recover X (up to homeomorphism) as E/G. Associated with each Steiner
system is its automorphism group, that is, the set of all..... The last statement of Lemma 2
yields the algebra isomorphism A = B. an orientation preserving homeomorphism a complete
set of representatives of the isomorphism classes of A-modules
most [see also: largely, extent, majority] There are at most two such r in (0, 1). The number
of distinct values that could be in a memory cell is at most s. Thus A is the union of B plus
an at most countable set. The two functions dier at most on a set of measure zero. If A
consists of at most one point, then..... The set F has the most points when..... In most cases
it turns out that..... Most measures that one meets are already complete. The proof shows
that if the points are drawn at random from the uniform distribution, most choices satisfy the
required bound. Although [1] deals mainly with the unit disc, most proofs are so constructed
that they apply to more general situations. Most of the theorems presented here have never
been published before. Most of this book is devoted to..... In the case where A is commutative,
as it will be in most of this paper, we have..... For most of the proof it suces to use the rough
bound p < 1. [Use most of before the, this, our etc.] A survey of the research on f
n
(x, y) up
to 1970 (most of it dealing with the case n = 1) was given in [3]. The proofs are, for the most
part, only sketched. This paper, for the most part, continues this line of investigation. We
shall be brief here and just estimate the terms that are the most challenging. Most probably,
his method will prove useful in..... What most interests us is whether.....
-most the rightmost expression [= the last one on the right] the innermost integral
mostly [see also: mainly, largely, primarily, part] Instead, we shall mostly work with the larger
class of..... Below, we shall be mostly interested in Poisson cohomology of the algebra V .
motivate [see also: inspire, prompt] This motivated Galvin to ask the following question. The
lemma motivates our calling R a generalized Picard group. It seems that the relations between
these concepts emerge most clearly when the setting is quite abstract, and this (rather than a
desire for mere generality) motivates our approach to the subject. Whether or not this is correct
does not matter; we are trying to motivate the proof that follows. We motivate our work in
Lipschitz domains by rst considering a model case, that of harmonic functions dened on the
upper half-space. The following theorem is clearly motivated by the classical LP-decomposition.
The present paper is motivated by the desire to make the subject as accessible as possible.
Motivated by (4), we obtain the following characterization of..... This motivated the second
author to introduce the notion of.....
motivation [see also: impulse, inspiration] Further motivation for looking at ideal class groups
comes from the eld of cryptography. The motivation for writing this paper was twofold. The
motivation for the results of this section is the following result of John (paraphrased slightly to
suit our purposes). This is the motivation behind the following process.
move [see also: go, proceed, carry, transfer] The problem is to move all the discs to the third
peg by moving only one at a time. The point A can be reached from B by moving along an edge
MATHEMATICAL ENGLISH USAGE 91
of G. Any point not in B is moved by f a distance equal to twice the distance to M. Part of
the conclusion is that F moves each z closer to the origin than it was. In the present paper we
move outside the random walk case and treat time-inhomogeneous convolutions. As the point
z moves around the unit circle, the corresponding J
z
s are rotations of angle t(z). Schenzels
formula frequently allows us to move back and forth between the commutative algebras of k[P]
and the combinatorics of P. We now move on to the question of local normal forms.
much [see also: extent, far, abundance, wealth, majority] Thus F is very much larger than G.
Although the denition may seem articial, it is actually very much in the spirit of Darbos
old argument in [5]. This interval is much smaller than that suggested by (8). Much less is
known about hyperbolically convex functions. The map G can be handled in much the same
way. We can multiply two elements of E by concatenating paths, much as in the denition
of the fundamental group. The strategy is much the same as for the proof of Theorem 2.
Much of the rest of the paper is devoted to..... How much of the foregoing can be extended to
the noncompact case? Another topic of great interest is how much of adjunction theory holds
for ample vector bundles. Let us now take a quick look at the class N, with the purpose of
determining how much of Theorems 1 and 2 is true here.
multiple Thus F is at most a multiple of G plus..... The last term is bounded by a constant
multiple of the norm of g.
multiplication The induced homomorphism is multiplication by 2. Finally, multiplication
by a permutation matrix will get the exponents in descending order. The Fourier transform
converts multiplication by a character into translation, and vice versa, it converts convolutions
to pointwise products.
multiply Multiplying through by f and integrating shows that..... Then A = B, as one sees by
multiplying out the product on the right. The theory of elementary divisors now shows that
by pre- and post-multiplying x by suitable elements of K we can reduce x to a diagonal matrix.
must [see also: have to, necessarily, force] Any algorithm to nd max must do at least n
comparisons. We must have Lf = 0, for otherwise we can replace f by f Lf. Our present
assumption implies that the last inequality in (8) must actually be an equality. If there are
to be any nontrivial solutions x then any odd prime must satisfy..... In outline, the argument
follows that of the single-valued setting, but there are several signicant issues that must be
addressed in the n-valued case. Nevertheless, in interpreting this conclusion, caution must be
exercised because the number of potential exceptions is huge. Theorem 3 may be interpreted
as saying that A = B, but it must then be remembered that.....
mutatis mutandis [= with necessary changes] We shall suppose that Assumptions 2.1 hold
mutatis mutandis.
N
name [see also: call, designate, term] 1 The name Riesz theorem is sometimes given to the
theorem which asserts that..... The name of Harald Bohr is attached to bG in recognition of his
work on almost periodic functions. This class of Hurewicz spaces has appeared under dierent
names in the study of small sets of reals.
2 This has numerous important consequences; to name just two, Brauer proved that..... Here
D is induced by the same named dierential on the algebra V .
92 JERZY TRZECIAK
namely Here D
0
and D
1
are discs with the same centre, namely b. The object of this paper is
to obtain improvements in two cases, namely for forms of degree 7 and 11. It has some basic
properties in common with another most important class of functions, namely, the continuous
ones. There is another, entirely dierent, way to see that A = B. Namely, one can rst show
that..... [Note the dierence between namely and that is: while namely introduces specic or
extra information, that is (or i.e.) introduces another way of putting what has already been
said.]
narrowly The answer depends on how broadly or narrowly the term matrix method is dened.
natural It is therefore natural to allow (5) to fail when x is not a continuity point of F. It is
natural to make the following denition. In this case it is natural (and notationally convenient)
to pick W to be..... A natural question to ask is how the quantities A(S, T) and B(S, T) vary
as S and T change.
naturally This space is naturally isomorphic to B. The question naturally arises whether
this representation is unique. Where we could, we have chosen these examples from naturally
occurring mathematical structures. [Note the double r in occurring.]
nature [see also: essence, substance, point] The two examples, E
1
and E
2
, dier by only a
single sequence, e, and they serve to illustrate the delicate nature of Theorem 2.
near [see also: close] Thus F is near 1. [Or: near to 1] for p near enough to q Denote by q
the point of A nearest to p. Here, of course, the set A produced is rather thin and certainly
nowhere near the densities we are looking for.
nearby We can write nearby points y as y = x +r.
nearly [see also: almost, practically] This is nearly the same as formula (6) of [7], which we can
recover by multiplying (2.3) by F(s).
neat [see also: nice, precise] The surprisingly neat answer is that the functions (2) span C(I) if
and only if G(a) = 0.
necessarily [see also: need, entail] For general rings, Out(R) is not necessarily well-behaved.
Clearly, F is bounded but it is not necessarily so after division by G. .....where P(d) denotes
the space of (not necessarily monic) polynomial functions of degree d. Necessarily, one of X
and Y is in Z. His proof is unnecessarily complicated.
necessary [see also: condition, need] We can assume, by decreasing n if necessary, that.....
Passing to a subsequence if necessary, we can assume that..... We can assume that p is as
close to q as is necessary for the following proof to work. This is necessary for determining
the constants in Theorem 2. Although these proofs run along similar lines, there are subtle
adjustments necessary to t the argument to each new situation. It is therefore unnecessary
to specify G on M.
necessitate [see also: entail, involve, force, require] Apart from a few embellishments necessi-
tated by some technical diculties, the ideas dier very little from those used to prove Lemma 4.
necessity [see also: need] We now prove necessity. [Or: the necessity]
need 1 [see also: necessity, require, necessary] By subtracting a constant if need be, we may
assume..... There is no need for the assumption that..... However, with the recent advent of
MATHEMATICAL ENGLISH USAGE 93
simulation based inference, the need for analytically tractable posteriors is no longer critical.
Hochberg pointed to the need for procedures that are more powerful than classical comparison
methods. Let us digress to locate the need for relative ergodicity in the above arguments.
2 [see also: require, necessitate, necessary] However, (9) needs handling with greater care. The
others being obvious, only (iv) needs proof. A drawback to Polya trees, and perhaps the main
reason why they have not seen much application within the Bayesian nonparametric literature,
is that an arbitrary partition tree needs to be specied. We need to make the smallest possible
jump. To recover Wieners famous result that Brownian paths are continuous, one needs to
use more sophisticated reasoning. This will help us nd what conditions on A are needed for
T(A) to be analytic. For F = R, no integration over M is needed in (5). The only property
of M needed is local compactness. More specialized notions from Banach space theory will be
introduced as needed.
3 [modal verb, in questions, negatives or with only; see also: have to] We need only consider
three cases. [Or: We only need to consider] Only shifts with k = 1 need be considered. It
turns out that nothing more need be done to obtain..... In that case Y need only be metrizable
(rather than completely metrizable). Then F need not satisfy (2). [= F does not necessarily
satisfy (2).] However, it need not be the case that V > W, as we shall see in the following
example.
negative [see also: positive, minus] Theorem 1 provides an answer in the negative. The only
references known to the authors are [A] and [V], where the case A = L(E) is settled in the
negative. the negative of a [= a]
neglect [see also: disregard, ignore] Clearly, the contribution from those r with A(r) > 0 can
be neglected. However, we prefer to avoid this issue altogether by neglecting the contribution
of B to S.
negligible We show that A is negligible compared with B.
neither Neither (1) nor (2) alone is sucient for (3) to hold. Thus A is neither symmetric
nor positive. Both X and Y are countable, but neither is nite. Neither of them is nite.
[Use neither when there are two alternatives; if there are more, use none.] Let u and v be
two distributions neither of which has compact support. As shown in Figure 3, neither curve
intersects X. In neither case can f be smooth. [Note the inversion after the negative clause.]
Both proofs are easy, so we give neither. Thus X is not nite; neither nor) is Y . Neither
is the problem simplied by assuming f = g.
never In representation theory, there can never be a B-map whose domain is nite-dimensional.
Most of the theorems presented here have never been published before. If one studies the
proof of..... it is apparent that (2) is never used. If the boundary is never hit then x
t
is a Feller
process under reasonable continuity assumptions.
nevertheless [see also: however, yet] Nevertheless, it might be possible to make sense of (2)
even for non-injective V by considering a multi-valued operator Z. We show that nevertheless
a positive proportion of the polynomials B
n
(x) satisfy Eisensteins criterion.
next [see also: subsequently, then] The next three comparisons are:..... We next show that.....
Next, (1) shows that (2) holds whenever g = f(n) for some n. the next-to-last inequality
nice [see also: neat] Neighbourhoods of points in these spaces appear at rst glance to have
94 JERZY TRZECIAK
a nice regular structure, but upon closer scrutiny, one sees that many neighbourhoods contain
collections of arcs hopelessly folded up. In the case of M
n
, the corresponding symmetric group
action on cohomology is particularly nice. Homogeneous spaces are, in a sense, the nicest
examples of Riemannian manifolds.
nicety It is clear that (up to set-theoretic niceties) this denes a partial order on the class of
R-equivalence classes of Borel maps out of the given space X.
no This set has no fewer elements than K has. Then X has no fewer than twenty elements.
The case n = 2 is of no interest since..... For F = R, no integration over M is needed in (5).
Thus F has no pole in U (hence none in V ). No x has more than one inverse. There is no
map such that..... For no x does the limit exist. [Note the inversion after the negative clause.]
No two members of A have an element in common. We conclude that, no matter what the
class of b is, we have an upper bound on M.
non- [Write consistently either nontrivial or non-trivial etc.; in some cases the hyphen is required:
a non-Euclidean metric, a non-locally convex space; never write non as a separate word.] for
any noncompact set [Not: for any not compact set; you can only use not predicatively, e.g.
for any set that is not compact]
none [see also: nothing] Thus F has no pole in U (hence none in V ). .....where none of the
sums is of the form..... Let A
n
be a sequence of positive integers none of which is 1 less than a
power of two. If there is an f..... then..... If there are is) none, we dene..... It follows that
the semigroup S
t
is none other than e(t)T.
nor Neither (1) nor (2) alone is sucient for (3) to hold. Clearly, A is neither symmetric nor
positive. However, X is not nite, nor is Y countable. [Note the inversion.] This topology is
compact, but not usually Hausdor, nor even T
1
.
norm Then F
n
tends to F in norm in the L
p
norm).
normally We will not normally specify the index set I.
not For m not an integer, the norm can be dened by interpolation. For f not in B,.....
However, M is generally not a manifold. It is not generally possible to restrict f to the
class D. We note that H is in fact not monotone if this condition is violated. [Do not use not
to negate an adjective placed before a noun. Write: a non-monotone function, and not: a not
monotone function.] Is there a relation between A and B? There certainly is not if..... Now
f is independent of the choice of (although the integral itself is not). Examples 1 and 2 give
two operators, the former bounded and the latter not, with..... Let us stress that c is a term
and not a subset of C. Obviously, S may be P, but it may also not be P. The result above
seems not to be a consequence of previous results. The ordered pair (a, b) can be chosen in 16
ways so as not to be a multiple of (c, d). Why not increase the precision of these statements?
notably [see also: especially, particular, particularly] Section 2 contains some specic prepara-
tory material, notably a brief discussion of the category of virtual vector bundles. [= the most
important part of it being a brief discussion etc.]
notation [see also: symbol] By abuse of notation, we continue to write f for f
1
. For simplicity,
we suppress the explicit dependence on x in the notation. For ease of notation [= To ease
notation], set I = I
f
. If G is clear from context, then we suppress reference to it in the
MATHEMATICAL ENGLISH USAGE 95
notation. This is in agreement with our previous notation. In this chapter we shall depart
from the previous notation and use the letter m not for Lebesgue measure, but for Lebesgue
measure divided by (2)
1/2
. Standard Banach space notation is used throughout. For clarity,
however, we record the notation that is used most heavily. The reader is cautioned that our
notation is in conict with that of [3]. Unfortunately, the notation from number theory slightly
conicts with the notation from probability theory. Any other unexplained notation is as
found in Fox (1995). The notation F < G will mean that..... Let us introduce the temporary
notation Ff for gfg
1
. One more piece of notation: throughout the paper we write..... for.....
From (ii), with an obvious change in notation, we get..... We do not use (5) in our proofs,
because it makes the notation more cumbersome.
note [see also: notice, observe, see, remark] Note that (3) is merely an abbreviation for the
statement that..... Perhaps it is appropriate at this point to note that a representing measure
is countably additive if and only if..... Part (b) follows from (a) on noting that A = B under the
conditions stated. Before going to the proof, it is worth noting that..... ....., as noted as was
noted) in Section 2. [Not: as it was noted] As noted before, there exists N homeomorphic
to P such that..... It should be noted that..... With the exception noted below, we follow
Stanleys presentation [3, Sec. 2].
nothing [see also: none] If n = 1, there is nothing to prove. Even in the case n = 2, the
application of Theorem 6 gives essentially nothing better than the inequality..... It turns out
that nothing more need be done to obtain..... However, (ii) is nothing but the statement
that..... [= only the statement that] The identity p(A) = 0 is nothing other than the Cayley-
Hamilton theorem. If nothing else, I hope to convince my readers that Segals theorem deserves
recognition as a profound contribution to Gaussian analysis.
notice [see also: note, observe, remark] It is important to notice some of the weaknesses inherent
in the above approach. In 1988, while attempting to generalize this result, the second author
noticed that..... In our next theorem, we state a characterization of..... which does not seem
to have been noticed previously.
notion [see also: concept, idea] We describe how the notion of positivity relates to the other
properties. How are these two optimality notions related? In general, they are not. This is
where the notion of an upper gradient comes in. Although standard, the notion of a virtual
vector bundle is not particularly well known. There is a fourth notion of phantom map which
bears the same relation to the third denition as the rst does to the second. Our notion of
nonanalytic integral encompasses such well known examples as..... More specialized notions
from Banach space theory will be introduced as needed. The benet of formulating our notion
of isomorphism section as above will become clear shortly. Are the notions of boundedly
approximately amenable and approximately amenable the same?
novelty Since most of the results presented are quite classical (the novelty lies in the arrangement,
and some of the proofs are new), I have not attempted to document the source of every item.
now The result will now be derived computationally. Moreras theorem shows now that f is
holomorphic. Now that we have the above claim, we can select..... Now choose a cycle c in
M as in Theorem 2.
nowhere a nowhere vanishing vector eld
96 JERZY TRZECIAK
number 1 [see also: abound, many, numerous, profusion, several, variety] Let q be the maximum
number of variables occurring in..... [Note the double r in occurring.] The total number of
vectors of this type is 36. the number of elements of X the number of terms in the sequence
(a
n
) such that..... Nevertheless, in interpreting this conclusion, caution must be exercised
because the number of potential exceptions is huge. For size 1 this makes no dierence, but
for sizes 2 and 3 it considerably cuts down the number of cases to be considered. All but a
nite number of the G
s
are empty. for all n except a nite number There are a large number
of examples showing that..... [Not: There is a large number] There are quite a number
of cases, but they can be described reasonably systematically. We give a number of results
concerning..... [= several] There are a number of equivalent denitions of the regular set.
2 The elements of D can be so numbered that..... They are numbered in order of increasing
diameter. Number the successive segments of the boundary line between A and B (marked
thickly in the picture) with the numbers 0, 1, ..., n, starting at the bottom. numbered in Arabic
numerals These from G number 45. [= there are 45 of them] The elements of G, numbering
122 in all, range from 9 to 2000.
numerous [see also: plentiful, abound, many, number, profusion] There are numerous results
in the literature relating spectral conditions to invertibility of f. This has numerous important
consequences; to name just two, Brauer proved that.....
O
obey [see also: satisfy] We rst check that t

obeys the condition for f(t). One easily checks

that whenever a sequence a
n
obeys the uniform bound a
n
< C, one has..... It follows that any
itinerary that obeys these four rules corresponds to a point in B.
object [see also: aim, purpose] The object of this section is to classify all the indecomposable
E-modules. The object of this paper is to obtain improvements in two cases, namely for forms
of degree 7 and 11.
objective [see also: aim, idea, purpose, intention, task] This achieves our objective of describ-
ing..... The objective is to choose a control u so as to minimize..... the key main/major/
primary/principal/broad/ultimate/stated) objective
obscure However, the connection with Gromovs work has been obscured in recent years by an
emphasis (in the algebraic topology literature) on conguration spaces.
observation [see also: remark, fact] The preceding observation, when looked at from a more
general point of view, leads to..... The key fundamental) observation is that if..... We
start with the observation that..... We shall have a number of opportunities in the following
arguments to make use of this simple observation. This simple observation lies at the heart of
the proof of our rst theorem.
observe [see also: note, notice, remark] The observed values of X will on average cluster around
points where..... Observe how the completeness of L
2
was used to guarantee the existence of f.
obstacle [see also: obstruction, prevent] The obstacle we have is that we must control the area
of G. It might seem at rst that the only obstacle is the fact that the group is not compact.
There are two obstacles to this approach. It is possible that the methods of this paper could
be used to....., but there remain considerable obstacles to overcome. We are able to surmount
this obstacle by an error analysis of the rate of convergence.
MATHEMATICAL ENGLISH USAGE 97
obstruction [see also: obstacle, prevent] But the T
n
need not be contractions in L
1
, which is
the main obstruction to applying standard arguments for densities.
obtain [see also: get, derive, gain] Theorem 2, at the end of Section 2, was not originally
obtained in the manner indicated there. A key step in obtaining (6) is Jensens inequality.
We thus obtain f = g. [Not: We obtain that f = g, We receive f = g.] Also, wherever
possible, we work with integer coecients, enabling us to obtain information about torsion.
See [17] for a brief account of the results obtained. [Not: the obtained results]
obvious [see also: clear, evident, apparent, plain] Now (8) makes it obvious that..... We regard
(1) as a mapping of S
2
into S
2
, with the obvious conventions concerning the point . We
shall also refer to a point as backward nonsingular, with the obvious analogous meaning. The
others being obvious, only (iv) needs proof.
obviously [see also: clearly, manifestly, apparently] Its restriction to M is obviously just f.
The left side of (3) obviously cannot decrease if r increases. Obviously, G
n
increases to G as
n .
occasion [see also: sometimes] We take advantage of this fact on several occasions, by not
actually specifying the topology under consideration. This connection has been exploited on
occasion [= sometimes] to construct various innite families of regular maps. This volume
was originally intended as a celebration volume marking the occasion of N. Wieners seventieth
birthday.
occasionally [see also: sometimes, occasion] The following variant of Theorem 2 is occasionally
useful.
occur [see also: arise, appear, emerge, happen, result] The only case requiring further analysis
occurs when f = 0. Both cases can occur. All inputs of size n are equally likely to occur.
Every nonincreasing function occurs as the rearrangement of some C(P) function. Equality
occurs in (9) if..... Where we could, we have chosen these examples from naturally occurring
mathematical structures. [Note the double r in occurring.]
occurrence Replace each occurrence of b by c. Throughout, C denotes a positive constant, not
necessarily the same at each occurrence.
odds [see also: chance, conict] The odds that he will win are ten to one. This is slightly at
odds [= inconsistent] with the terminology of [4], as Fox denes the trace lter to be the normal
lter generated by A.
of a vector of norm 1 a ball of radius r an element of nite order the group of invertible
elements of E Actually, S has the much stronger property of being convex. This method has
the disadvantage of not being intrinsic. However, only ve of these are distinct. The best
known of these is the Knaster continuum. Of these, (i) and (ii) are almost immediate from
the denition. There has since been a series of improvements, of which we briey mention
the work of Levinson. Suppose that of all such solutions, (x, y, z) is one with y minimal.
Theorem 2 of [8] statement (ii) of Proposition 7 However, this cannot be proved of the
cardinal function d(X). For example, F reaches a relative maximum of 5.2 at about x = 2.1.
The only additional feature is the appearance of a factor of 2.
o [see also: away, beyond, outside] a function with poles o K vanishing o K) an extension
98 JERZY TRZECIAK
of f o U The kernel is C
0
on X X o the diagonal. a Blaschke product with at least one
zero o the origin If we stay a xed distance o the critical line, we do not expect Benford
behaviour. We close this o by characterizing..... Let M

be the minor obtained by crossing

o the last row and column of M. Continuity then nishes o the argument. This can be
read o from (8).
oer [see also: give, provide, furnish, supply] We oer numerical evidence to support a conjecture
that there exist innitely many primes of this type. In this paper we oer some useful tools for
bounding d-pseudoprimes. We suspect that our methods should extend to oer at least some
description of characteristic factor-tuples for larger numbers of commuting transformations.
We close this paper by oering some questions and problems for further research.
oset [see also: cancel, compensate] This decrease is oset by the contribution from the poles.
Osetting the eect of the pole at t = 0 requires more work.
often [see also: frequently, repeatedly] It often does not matter whether..... This property
is of course already well known in many cases, often with little or no restrictions on V . The
parent and child relations often used with trees can be derived once a root is specied. The
question as to how often we should expect the class number to be divisible by p is also of some
interest. innitely often
old We shall need ways of constructing new triangulations from old ones which alter the f-vector
in a predictable fashion. This lends precision to an old assertion of Dini:..... Although the
denition may seem articial, it is actually very much in the spirit of Darbos old argument
in [5].
omission There are nite measures for which the omission of (iv) would make (b) false.
omit [see also: drop, delete, remove, dispense with] The conclusion does not follow if continuity
is omitted from the hypotheses. So far it seems not to be known whether the geometric
condition on X can be omitted. We omit further details. The wedge denotes that e
i
has
been omitted. However, if B were omitted in (1), the case n = 0 would imply Nf = 1, an
undesirable restriction. To simplify the writing, we take a = 0 and omit the subscripts a.
on The proof is by induction on n. For brevity, we drop the subscript t on h
t
. Here H is
a free R-module on as many generators as there are path components of X. the amount the
gambler wins on the nth play However, A clearly has a topology T on it, namely,..... On
making use of the bound..... we conclude that..... On combining this with our other estimates
in (3.5), we deduce that..... This bound is integrable and on integration one again obtains the
same form of..... On restriction to B we have a homomorphism f : B C. We establish our
results both unconditionally and on the assumption of the Riemann Hypothesis. The following
example shows that the degree of smoothness predicted by Theorem 6 cannot be improved on.
We now show the following improvement on (2). This is essentially a variation on the prime
number theorem. We will call on this version of the inverse theorem when we come to our
applications in Section 2. We now move on to the question of local normal forms. Then (5)
takes on the form..... [Or: takes the form]
once The parent and child relations often used with trees can be derived once a root is specied.
Once this is done, the proof continues thus:..... Once the dissipation relation is in hand, no
further work is required. This procedure, once implemented, can thereafter be applied with
MATHEMATICAL ENGLISH USAGE 99
great eectiveness. Moreover, it follows at once from (1) that..... Let us state once more, in
dierent words, what the preceding result says if p = 1.
one One more case merits mentioning here. One reason for using this rather elaborate model
is that it permits a simple and concrete denition of the realization. One simple way to get a
Banach algebra which is not amenable is to take..... Our denition agrees with the one in [3].
The cases p = 1 and p = 2 will be the ones of interest to us. Consider the dierences between
these integrals and the corresponding ones with f replaced by g. It has some basic properties
in common with another most important class of functions, namely, the continuous ones. The
geodesics (8) are the only ones that realize the distance between their endpoints. Then the
one and only integral curve of L starting from x is the straight line l. Then G has ten normal
subgroups and as many non-normal ones. The rst two are simpler than the third one. [Or:
the third; not: The rst two ones] Now, F has many points of continuity. Suppose x is one.
Each of the functions on the right of (9) is one to which our theorem applies. A principal
ideal is one that is generated by a single element. If f is an n-simplex in U, then f

is one
in V . Suppose that of all such solutions, (x, y, z) is one with y minimal. In addition to a
contribution to W
1
, there may also be one to W
2
. Necessarily, one of X and Y is in Z. Here
the interesting questions are not about individual examples, but about the asymptotic behaviour
of the set of examples as one or another of the invariants (such as the genus) goes to innity.
Instead of dealing with lines one by one, we deal with collections of lines simultaneously.
Clearly, A is thereby put in one-to-one correspondence with B. The algorithm examines only
roughly one-quarter to one-third of the characters. Asymptotically, more than one-fth of the
polynomials B
n
(x) are irreducible. The other player is one-third as fast. [The numbers 1 to
12, when used for counting objects (without units of measurement), should be written in words:
There exists exactly one such map; for other uses, sometimes gures permit avoiding ambiguity,
e.g. it is 1 less than the other.]
only [see also: alone, just, merely, single, solely] Assume that the only functions v, w satisfying
(2) are v = w = 0. Then the one and only integral curve of L starting from x is the straight
line l. Here x is the set whose only member is x. The problem is to move all the discs to
the third peg by moving only one at a time. However, only ve of these are distinct. Note
that F is dened only up to an additive constant. We need only consider the case..... [Or: We
only need to consider] The proof will only be indicated briey. We have to change the proof
of Lemma 3 only slightly. To prove (8), it only remains to verify that..... Thus X assumes
the values 0 and 1 only. Only for x = 1 does the limit exist. [Note the inversion.] If we know
a covering space E of X then not only do we know that..... but we can also recover X (up to
homeomorphism) as E/G.
onwards from line 6 onwards
open [see also: unresolved] These results leave open the basic case k = . Conjecture 2 of
[KH], to the eect that [=meaning that] there is no relation P with E(P) = 1, still remains
open.
operate [see also: act] When operated on by a rotation, each of these vectors is mapped to.....
opportunity We take this opportunity to correct a minor error in Lemma 2 of [PS]. We shall
have a number of opportunities in the following arguments to make use of this simple observation.
oppose [see also: contrary, contrast, unlike] As opposed to the situation considered in [5], the
100 JERZY TRZECIAK
functions used here are..... [Not: Contrary to the situation] The rest of the paper concerns
the existence of greedy bases (as opposed to basic sequences).
opposite [see also: reverse] For the opposite inclusion, suppose that..... the opposite inequality
Then D is the face of the simplex s opposite to A. [Or: opposite A] (see the page opposite)
or There is no recursive or denable R such that..... The answer depends on how broadly or
narrowly the term matrix method is dened. If..... then R is right Noetherian provided R is
semiprime [2] or commutative [4] or R/N has zero socle. The case when f is decreasing can
be proved similarly, or else can be deduced from..... Then f = g, or equivalently a(f) = a(g).
Here the interesting questions are not about individual examples, but about the asymptotic
behaviour of the set of examples as one or another of the invariants (such as the genus) goes to
innity. Moreover, for L tame or otherwise, it may happen that E is a free module. Then F
may or may not x B. The question we shall be concerned with is whether or not f is..... Its
role is to rule out having two or more consecutive P-moves. Any vector with three or fewer 1s
in the last twelve places has at least eight 1s in all. The intervals we are concerned with are
either completely inside A or completely inside B. By Corollary 2, distinct 8-sets have either
zero, two or four elements in common. Either f or g must be bounded. Any map either has
a xed point, or sends some point to its antipode.
order 1 [see also: arrangement, organization] Comparisons are done in left-to-right order.
They are numbered in order of increasing diameter. The diagram of L + M is obtained by
taking the rows of the diagrams of L and M and reassembling them in order of decreasing length.
in reverse order arranged in increasing order The interchange in the order of integration was
legitimate, since..... One is tempted to reverse the order of integrations but that is illegitimate
here. The rst 15 chapters should be taken up in the order in which they are presented, except
for Chapter 9, which may be postponed. an element of order a power of p an element of
prime power order Then F has continuous y-derivatives of all orders up to order k). Thus F
vanishes to order 3 to innite order) at x. Unfortunately, the details of the calculations were
omitted, and there is some doubt on whether the result is correct since our analysis suggests
that P
2
must vanish to third order; the presence of L
2
is also suspect. In all our analysis,
only the order of magnitude of P will be signicant. A form of the central limit theorem is
used to show that for large n,E(R
n
) is on the order of c

n, from which it follows that E(R

n
S
n
)
does not drop below the order of n
4/3
. First, some remarks concerning the term closure are
in order. In order to make our description of this process precise, we dene.....
2 [see also: arrange, organize] We partially order M by declaring X < Y to mean that.....
Let P be the set of all intervals in J ordered by inclusion.
organization [see also: order, arrangement, outline] The organization of the paper is the fol-
lowing.
organize [see also: arrange, order] The paper is organized as follows. Our presentation is
therefore organized in such a way that the analogies between the concepts of topological space
and continuous function, on the one hand, and of measurable space and measurable function,
on the other, are strongly emphasized.
origin [see also: source] The word singular as applied to measures has its origin in this phe-
nomenon. a ball of radius r about the origin
MATHEMATICAL ENGLISH USAGE 101
original [see also: rst] We now turn to Gromovs original result. This was one of the major
steps in Wieners original proof of his Tauberian theorem.
originally [see also: initially, beginning, rst] Theorem 2, at the end of Section 2, was not
originally obtained in the manner indicated there.
originate [see also: start, derive] This theory was originated by Gelfand in 1941. The argument
is now completed by means of techniques originated in the work of Stolz [3].
other [see also: dierent] Every region in the plane (other than the plane itself) is conformally
equivalent to U. Clearly, A

weights are sharp weights. That there are no others is the main
result of Section 2. Then V has the following invariant subspaces, and no others:..... The
others being obvious, only (iv) needs proof. It follows that the semigroup S
t
is none other
than e(t)T. The identity p(A) = 0 is nothing other than the Cayley-Hamilton theorem. Each
vertex is adjacent to q others. The other inequality is just as easy to prove. [= the other of the
two mentioned] the other end of the interval In this and the other theorems of this section,
the X
n
are any independent random variables with a common distribution. One of these lies
in the union of the other two. On combining this with our other estimates in (3.5), we deduce
that..... Any other unexplained notation is as found in Fox (1995). The next corollary shows
among other things that..... [Not: among others] Our result generalizes Urysohns extension
theorems, among others. [= among other theorems] These n disjoint boxes are translates of
each other.
otherwise [see also: contrary] We must have Lf = 0, for otherwise we can replace f by f Lf.
We claim that f(z) > 1. Otherwise, the disc D would intersect B. We now prove..... Indeed,
suppose otherwise. Then..... Unless otherwise stated, we assume that..... Moreover, for L
tame or otherwise, it may happen that E is a free module. Simplicity (or otherwise) of the
underlying graphs will be discussed in the next section.
ought There ought therefore to be a point x such that.....
out Choose one out of these ten. in nine cases out of ten Dene a
k
to be the probability that
exactly k out of the 2n values X
i
exceed T, conditional on X
0
> T. The only edges out of 3
lead to 2 or into B. It is clear that (up to set-theoretic niceties) this denes a partial order on
the class of R-equivalence classes of Borel maps out of the given space X. A second technique
for creating new triangulations out of old ones is central retriangulation. The detailed analysis
of..... is carried out in Section 2. This term drops out when f is dierentiated. We were
surprised to nd out that..... at nding out that.....) This accords with the intuition that
as we pass down the coding tree, we nd out more and more detailed information about the
ordering actually represented. Our study grew out of some valuable conversations with Kirk
Douglas. We lay out the details of this generalization in the rst part of this paper. The
image of U under f misses out more than three points of the sphere. Then A = B, as one
sees by multiplying out the product on the right. One unusual feature of the solution should
be pointed out. To round out the picture presented by Theorem 5, we mention the following
consequence of..... The possibility A = B is ruled out in the same way. By modifying the
technique set out [= presented] in [3], we obtain..... With this denition of a tree, no vertex
is singled out as the root. It turns out that these properties play no role in the proof. By
writing out the appropriate equations, we see that this is equivalent to.....
outcome [see also: eect, result] At rst glance Lemma 2 seems to yield four possible outcomes.
102 JERZY TRZECIAK
It is impossible to predict the eventual outcome of the process. It produces the same outcome
whichever path is taken. [= no matter which] We tabulate the outcome for n 10; in
particular, the column headed R lists R(n) truncated to three decimal places.
outline [see also: sketch, organization] 1 In brief outline, here is the main idea of the proof. In
outline, the argument follows that of the single-valued setting, but there are several signicant
issues that must be addressed in the n-valued case. The outline of this paper is as follows.
2 We shall briey outline the necessary changes. Where it is possible, we outline the proofs so
that the reader will not have to hunt for obscure references.
outnumber In that population, women outnumbered men by) 2 to 1.
output The algorithm outputs a list of.....
outset [see also: beginning, start] We remark at the outset that this formula makes sense,
because..... It seems preferable, for claritys sake, not to present the construction at the outset
in the greatest generality possible.
outside [see also: beyond, away, o] The family of 4-sets will be used to generate a symmetry
outside N but in M. Outside the xed points, the group T acts freely on M. In the latter case
we may simply adjust F to equal 1 on the Borel set where it falls outside the specied interval.
In the present paper we move outside the random walk case and treat time-inhomogeneous
convolutions.
outstanding an outstanding scientist [= excellent] Two tasks still remain outstanding. [= not
done]
outward(s) It is sucient to prove that this vector eld points outwards on M. [In most
adverbial uses, outward and outwards are interchangeable; as an adjective, the more standard
form is outward: the outward layer.]
over Note that F is the product over the integers m in B. Hence V (x) is the maximum value of
J
x
(v) over all controls v. .....where E runs over the family B. To show the greater simplicity
of our method over Browns, let us..... Over the past ten years the isomorphic structure of
spaces of weighted holomorphic functions has been largely [= almost completely] determined.
Firms employing over 1000 people accounted for 50% of total employment. [= represented 50%]
The method of proof carries over to domains satisfying..... The equation PK = 0 then goes
over to QK = 0. Here s takes over the role of the time parameter.
overall Now (c) asserts only that the overall maximum of f on U is attained at some point of
the boundary.
overcome [see also: circumvent, get around] To overcome this problem, we revise our denition
of a branch. It is possible that the methods of this paper could be used to....., but there remain
considerable obstacles to overcome.
overlap The two categories overlap to some extent. non-overlapping intervals
overlook [see also: miss] Our rst result generalizes (8) by exploiting some general facts seem-
ingly overlooked by the aforementioned authors.
overview [see also: survey] We start with a brief overview of our strategy. Section 2 contains
an overview of the necessary background.
MATHEMATICAL ENGLISH USAGE 103
owe The present paper owes a great debt to Strangs work. The exposition owes much to the
work of Dold described in [3].
own Theorem 2 is interesting in its own right.
P
page on page 13 [Not: on the page 13] at the top of page 4 (see the last paragraph but
one of page 24) (see the page opposite) Our method of proof will be an adaptation of the
reasoning used on pp. 7172 of [3].
pair 1 The ordered pair (a, b) can be chosen in 16 ways so as not to be a multiple of (c, d).
There cannot be two edges between one pair of vertices.
2 Write out the integers from 1 to n. Pair up the rst and the last, the second and next to last,
etc.
paper [see also: article] in paper [3] [Or: in the paper [3]; better: in [3]] in a companion
paper [4] The aim of this paper is to bring together two areas in which..... In the present
paper we move outside the random walk case and treat time-inhomogeneous convolutions.
There are, however, a few important papers of which we were unaware until fairly recently.
These volumes bring together all of R. Bings published mathematical papers.
paragraph [indented fragment of text; ,= section] Let the notation be as in the preceding
paragraph. (see the last paragraph but one of page 24)
parallel 1 The proof runs parallel to that of Lemma 2. [Not: parallelly]
2 From the viewpoint of the Fox theorem, there is not an exact parallel between the odds and
the evens.
3 The proof closely parallels that of Theorem 1.
parameter The parameter interval was here taken to be (0, 1). Here s takes over the role of
the time parameter.
parametrize [= parameterize] a curve parametrized on the interval [0, 1]
paraphrase The motivation for the results of this section is the following result of John (para-
phrased slightly to suit our purposes).
parenthesis [pl. parentheses; see also: bracket] With no parentheses, and take priority
over the group operation. the expression in parentheses
part [see also: role] This is part of a larger project to study the Galois groups of periodic points
of arbitrary polynomial maps. [Or: This is a part] Part of the conclusion is that F moves
each z closer to the origin than it was. The equality f = g, which is part of Theorem 2,
implies..... This is the hard part of Joness theorem. The if part is straightforward.
Translations have already played an important part in our study of Fourier transforms. The
key part is to show that the submanifolds U
k
t together to form a complex submanifold.
Mary Lane deserves our special thanks for her part in bringing this volume to a successful
completion. Denition 3 is motivated in part by certain dierential operators to be introduced
in Section 3. This paper, for the most part, continues this line of investigation. the greater
signicant/substantial/principal/central/indispensable/integral/easy) part
104 JERZY TRZECIAK
partial As a partial answer to that question, we prove:..... However, it is not known when L(S)
is weakly amenable; for some partial results, see [BS].
partially [= partly] Research partially supported by NSF Grant No. 23456.
particular [see also: specic, concrete, notably] When it is necessary to emphasize one particu-
lar coordinate, we write..... The probability that any particular edge is a bond is 2p. Rather
than discuss this in full generality, let us look at a particular situation of this kind. Here the
constants of proportionality depend on the particular curve being considered. One might hope
that in the particular case of the GL energy, this could be established, but we do not see an easy
path to such a conclusion. In particular, E contains a copy of l
1
. So we must in particular
show that sets like this are not added.
particularly [see also: especially] This realization is particularly convenient for determining.....
Although standard, the notion of a virtual vector bundle is not particularly well known. We
make the following provisional denition, which is neither general nor particularly elegant, but
is convenient for the induction which is to follow. There is also a decrease in clustering as n
increases, particularly for normally distributed X.
partition 1 Intuitively, entropy of a partition is a measure of its information contentthe larger
the entropy, the larger the information content. a partition of unity subordinate to the covering
U
i

2 We can partition [0, 1] into n intervals by taking.....

pass [see also: by-product, incidentally, turn] This enables us to pass from a compact group G
to a maximal torus T. [Or: enables passing, not: enables to pass] Passing to a subsequence if
necessary, we can assume that..... Before passing to 4(b), we observe that..... Every path on G
passes through vertices of V and W alternately. We note in passing that Fox has subsequently
improved Barness result by showing that..... It seems appropriate to mention in passing the
corresponding formula for the cohomology of B
n
.
passage [see also: transfer, transition] The passage from bounded f to general f is easily
eected. The properties of A are preserved in the passage from V to C(V ). Now (1) follows
after passage to the limit as n .
path One might hope that in the particular case of the GL energy, this could be established, but
we do not see an easy path to such a conclusion. a path starting terminating) at x a path
obtained by going from A to B along the lower half of the circle
pattern 1 [see also: model, scheme] Other types t into this pattern as well. The proof that
a < b follows the standard pattern. Thus, although we follow the general pattern of proof of
Theorem A, we must also introduce new ideas to deal with the lack of product structure.
2 [see also: model] The proof is patterned upon after) Section 2 of [6].
peculiar [see also: characteristic] We base our development on two properties of prolongation
peculiar to this case. [= only found in this case]
penultimate [= next to last] The penultimate inequality is justied by the fact that.....
Then....., where the fact that A = B was used on the penultimate line.
percent [= per cent] About 40 percent of the solar energy is concentrated in the region.....
an increase of 5% a 5% increase) in the cost of living The quantity A was greater by a mere
MATHEMATICAL ENGLISH USAGE 105
20%. Quarterly sales rose by 9% compared with the same period in 2007. % increase on
previous year [in a diagram caption] Firms employing over 1000 people accounted for 50% of
total employment. [= represented 50%] Subsidies on these commodities total 25 per cent of
the budget.
percentage A positive percentage of summands occur in all the k partitions. Selberg [Sel] was
the rst to prove that a positive percentage of zeros lie on the critical line. A small percentage
reduction in the cost of materials resulted in a signicant increase in prot.
perform [see also: carry out, make] These calculations can be performed entirely in terms of
the generators of G. Its role is to rule out having two or more consecutive P-moves (on the
grounds that they can be performed in one go).
perhaps [see also: possibly] After, perhaps, passing to a subsequence, we have..... Perhaps
it is appropriate at this point to note that a representing measure is countably additive if and
only if..... Perhaps the most important problem involving f-vectors is whether or not Johns
conditions extend to spheres. A drawback to Polya trees, and perhaps the main reason why
they have not seen much application within the Bayesian nonparametric literature, is that an
arbitrary partition tree needs to be specied.
period Each period 3 orbit of F leads to ve period 2 orbits of F
3
. a map of period p in t
periodic a 2-periodic function
permission The nal lemma is due to F. Black and is included with his kind permission.
permit [sth; sb to do sth; see also: admit, allow, enable] This will permit us to demonstrate
that..... [Not: permit to demonstrate] This permits the extension of boundary estimates to
systems of type II. One reason for using this rather elaborate model is that it permits a simple
and concrete denition of the realization. Section 6 contains a formula which permits transfer
of the results in Section 2 to sums of independent random variables. Note that m is permitted
to vary with the number of inputs. Here the right side is permitted to be innite. In contrast
to [1], we permit the metric to be incomplete.
permute If a, b, and c are permuted cyclically, the left side of (2) is unaected.
persist [see also: continue, remain] This inequality persists as r 1. [= continues to hold]
A substantial no-mans-land persists in which neither alternative has been proved. This leaves
the hope that a ratio theorem may persist in a more general setting.
perspective [see also: view, viewpoint] This puts a completely dierent perspective on Foxs
results. Those results permit us to view this fragment of the theory from a broader perspective.
A change in perspective allows us to gain not only more general, but also ner results than
in [ST]. a new wider/proper/right/wrong/theoretical) perspective
pertain [to sth; = be related or relevant to sth] The paper provides a proof of a combinatorial
result that pertains to the characterization of..... The author gratefully acknowledges the
referees helpful comments pertaining to the rst draft of this paper.
pertinent [see also: relevant, connect] He proved the following theorem (see Section 2 for
pertinent denitions).
106 JERZY TRZECIAK
phenomenon [pl. phenomena] Formula (7) exhibits the same phenomenon. This rather coun-
terintuitive phenomenon is exhibited by the following example. This phenomenon is called
overconvergence. The word singular as applied to measures has its origin in this phenomenon.
a universal isolated/curious/remarkable/complex) phenomenon
phrase [see also: reformulate, rephrase, restate] Here is an alternative phrasing of part (1):.....
pick Pick the rst arc of length 1 in this sequence. We can continue to pick elements of B as
above. But there are only nitely many such, a contradiction.
picture 1 In Section 4 we worked out a fairly detailed picture of linear H-systems. To round
out the picture presented by Theorem 5, we mention the following consequence of..... The skew
diagram L-M is the shaded region in the picture below.
2 [see also: represent] This is easy to picture geometrically because.....
piece 1 We now start piling the pieces on top of each other. One more piece of notation:
throughout the paper we write..... for.....
2 [see also: combine, gather] The liftings on A and B agree on AB, hence we can piece them
together to obtain..... Piecing the above observations together, we can now show that.....
place 1 [see also: location, position, replace, instead] .....where 1 appears in at) the nth place
Any vector with three or fewer 1s in the last twelve places has at least eight 1s in all. Values
computed for the right side of (2) were rounded up in the fourth decimal place. We tabulate the
outcome for n 10; in particular, the column headed R lists R(n) truncated to three decimal
places. Apply this to f in place of g to obtain..... Then the conclusion holds with P-cell in
place of cell. All of the action in creating S
i+1
takes place in the individual cells of type 2
or 3. Moreover, one has estimates on the rate at which this convergence is taking place.
2 [see also: put, insert, impose] Some restrictions must be placed on the behaviour of f.
To place Theorem 1 in context, consider two real vector elds..... In this section we gather
some miscellaneous results that are more or less standard. These will be used to calculate the
constants c

, and we place them here to avoid interrupting the forthcoming arguments.

plain [see also: clear, obvious, evident, apparent] But this last assertion follows from Corollary 2,
it being plain that..... [= because it is plain]
plausible [= very likely; see also: possible] It seems plausible that..... but we have been able to
establish this only in certain cases.
play 1 Brucks theorem on common xed points for commuting nonexpansive mappings is then
brought into play by noting that.....
2 It turns out that these properties play no role in the proof. In closing this section we take up
a result which will play a pivotal role in the characterization of..... Translations have already
played an important part in our study of Fourier transforms.
plentiful [see also: numerous, abound, profusion, variety, abundance] On the other hand, if d
is a square, then Lucas primes appear to be plentiful.
plug [see also: insert, put] After plugging this estimate into (2), we conclude that.....
plus Thus A is the union of B plus an at most countable set. Therefore F is at most a multiple
of G plus a contribution from.....
MATHEMATICAL ENGLISH USAGE 107
point 1 [see also: issue, subject, matter, thing, topic, detail, aspect, feature, element, character-
istic, ingredient, essence, core] The preceding observation, when looked at from a more general
point of view, leads to..... It is this point of view which is close to that used in C

-algebras.
The point of the lemma is that it allows one to..... The point is that the operator is now
much easier to analyse than is the case in the original setting of the space B. An important
point is that..... The only point remaining concerns the behaviour of..... In [2], this theorem
is made the starting point of Gelfand theory. See [KT] for discussion of this technical point.
We shall not pursue this point here. We would like to know..... but that is beyond our reach
at this point. At this point, the reader is urged to review the denitions of..... Perhaps it is
appropriate at this point to note that a representing measure is countably additive if and only
if..... The arguments from this point up to Theorem 2 do not depend on.....
2 The vector v points in the negative direction points north) for x < 0. The vector points
outwards from M. It is sucient to prove that this vector eld points outwards on M. Here
M is a unit vector normal to M pointing into M. the inward pointing unit normal to M
All the evidence points to the validity of the conjecture. Hochberg pointed to the need for
procedures that are more powerful than classical comparison methods. One unusual feature of
the solution should be pointed out. However, as pointed out right after (5),.....
pose [see also: present, raise] We now pose a problem whose solution will aord an illustration
of how the Plancherel theorem can be used. The factor Gf poses no problem because..... [Do
not write: We pose f =..... if you mean: We set dene) f =.....]
position [see also: location, place] Then K appears in the last position in the list. The vector
v has at least n ones in its last m positions. This is the same as asking which row vectors in R
have diering entries at positions i and j. We are now in a position to prove our main result.
[Better: We can now prove] This puts us in a position to apply Lemma 2 to deduce that.....
After receiving his PhD he took a position at the University of Texas.
positive [see also: armative] This question was answered negatively in [5]. However, on the
positive side, Davies [5] proved that..... By allowing f to have both positive and negative
coecients, we obtain..... Here u
+
and u

are the positive and the negative parts of u, as

dened in Section 5. We show that nevertheless a positive proportion of the polynomials B
n
(x)
satisfy Eisensteins criterion. However, F is only nonnegative rather than strictly positive, as
one may have expected. Let Q denote the set of positive denite forms (including imprimitive
ones, if there are any).
positively What is F(c) if c is a positively oriented circle?
possession [see also: have, presence] Hence, the possession of Willard terms implies X = A.
possibility [see also: case] The possibility A = is not excluded. The only possibility for G
not to be 1-1 is that there exists..... The remaining possibility is that v is labelled 2. These
two possibilities cannot arise simultaneously. However, this still left open the possibility that
the converse of Theorem 3 held for m = 3. The reader may wonder why we have apparently
ignored the possibility of obtaining a better lower bound by considering..... [Not: the possibility
to obtain]
possible [see also: feasible, likely, plausible, enable] It is not generally possible to restrict f to
the class D. It is possible for the hull of J to have exactly two points. Is it possible to have
m(E) < 1 for such a set? the shortest possible way the maximum possible density This
108 JERZY TRZECIAK
result is best possible. [Or: the best possible] The above bound on a
n
is close to best possible.
We shall try to give it the simplest representation possible. It seems preferable, for claritys
sake, not to present the construction at the outset in the greatest generality possible. Every
possible such sequence gives rise to..... Then, for any two xed points that Wagners method
does not nd to be equivalent, he considers the possible lengths of potential solutions to (1). We
wish to arrange that f be as smooth as possible. For this purpose, it is necessary to understand
the mapping properties of B on as large a function space as possible. We follow where possible
the argument of Lang [9]. Also, wherever possible, we work with integer coecients, enabling
us to obtain information about torsion. This makes possible the proof of..... This makes it
possible to show that..... [Note that the it is necessary here.]
possibly [,= eventually; see also: perhaps, conceivably] The only points (z, w) at which the
continuity of g is possibly in doubt have z = 0. Clearly, F is invertible except possibly on
an at most countable set. We dene a (possibly unbounded) operator A by..... Although
individually these systems can still be quite complicated, a possibly more tractable task is to
describe their possible joint distributions. This yields (5) again (possibly with a dierent C).
We say that L is nitely aligned if A(L) is nite (possibly empty).
postpone [see also: defer] We postpone the proof until Section 2. The rst 15 chapters should
be taken up in the order in which they are presented, except for Chapter 9, which may be
postponed.
potential Then, for any two xed points that Wagners method does not nd to be equivalent,
he considers the possible lengths of potential solutions to (1).
power an element of prime power order of order a power of p) Raising this to the pth power
[= to the power p], we obtain..... Every prime in the factorization appears to an even power.
practically [see also: almost, nearly] However, in many applications the existence of such an R
is practically impossible to verify.
practice In addition to illustrating how our formulas work in practice, it provides a counterexam-
ple to Browns conjecture. In practice, D is usually too large a set to work with. [In British
English, practice is a noun, and practise is a verb; in American English, both spellings are used
for both noun and verb forms.]
preassign [see also: prescribe, give] Runges theorem will now be used to prove that meromor-
phic functions can be constructed with arbitrarily preassigned poles.
precede [see also: above, foregoing, previous] Let the notation be as in the preceding paragraph.
[Not: preceeding] The preceding observation, when looked at from a more general point of
view, leads to..... If A = B, how does the situation dier from the preceding one? Indeed, by
the preceding, we know that.....
precise [adjective, not verb; see also: accurate, rigorous, concrete, specic, very] This is made
more precise by the following denition. [Not: This is precised] It is important that the
orders of F and G are comparable, a statement made more precise by the following lemma.
To be precise, A is only left-continuous at 0. Put this way, the question is not precise enough.
Actually, the proof gives an even more precise conclusion:.....
precisely [see also: exactly, specically] We do not exclude the possibility that A consists
MATHEMATICAL ENGLISH USAGE 109
of precisely the polynomials. The resulting metric space consists precisely of the Lebesgue
integrable functions, provided we identify any two that are equal almost everywhere. Precisely
r of the intervals A
i
are closed. Thus A and B are at distance precisely d. We have d(f, g) = 0
precisely when f = g a.e. Important analytic dierences appear when one writes down precisely
what is meant by..... More precisely, f is just separately continuous.
precision This lends precision to an old assertion of Dini:.....
preclude [see also: prevent, rule out] Here, continuity precludes the existence of singularities.
predict It is impossible to predict the eventual outcome of the process. The following example
shows that the degree of smoothness predicted by Theorem 6 cannot be improved on.
predictable We shall need ways of constructing new triangulations from old ones which alter
the f-vector in a predictable fashion.
prefer [sth to sth] The reason for preferring (1) to (2) is simply that (1) is manifestly invariant.
[Note the double r in preferring.] However, we prefer to avoid this issue altogether by neglecting
the contribution of B to S.
preferable It seems preferable, for claritys sake, not to present the construction at the outset
in the greatest generality possible.
preliminary Following these preliminary remarks, we now state..... We give the important
preliminaries from descriptive set theory in Section 2.
preparatory Section 2 contains some specic preparatory material, notably a brief discussion of
the category of virtual vector bundles.
prerequisite The prerequisite for this book is a good course in advanced calculus.
prescribe [see also: preassign, give] The Taylor expansion of f at 0 can be prescribed up to
any nite order. In fact, we can do even better, and prescribe nitely many derivatives at each
point of A.
presence [see also: existence, possession] Unfortunately, because of the possible presence of
cusps, this need not be true. The main dierence from the case of nite coding trees is the
presence of limits. The presence here of the direct summand H is simply to prevent A from
having disconnected spectrum. Unfortunately, the details of the calculations were omitted, and
there is some doubt on whether the result is correct since our analysis suggests that P
2
must
vanish to third order; the presence of L
2
is also suspect. Sarkar discussed multiple testing in
the presence of positive dependence.
present 1 Note that any Euclidean space satises the present hypothesis. The present paper
owes a great debt to Strangs work. At present, we merely show how this extremal problem
can be used to characterize..... Inserting additional edges destroys no edges that were already
present. In the function eld case the poles at s = 0 and 1 are still present.
2 [see also: lay out, set out, represent] Pointwise convergence presents a more delicate problem.
The analogue of Theorem 1 presents no diculty. Proposition 2 presents examples of.....
Here we present another homomorphism that is of type MN. To round out the picture presented
by Theorem 5, we mention the following consequence of..... Since most of the results presented
are quite classical (the novelty lies in the arrangement, and some of the proofs are new), I have
not attempted to document the source of every item.
110 JERZY TRZECIAK
presentation [see also: exposition] The author thanks the referee for his helpful suggestions
concerning the presentation of this paper. Our presentation is therefore organized in such a way
that the analogies between the concepts of topological space and continuous function, on the one
hand, and of measurable space and measurable function, on the other, are strongly emphasized.
We note that the assumption of GCH is made for convenience and ease of presentation. With
the exception noted below, we follow Stanleys presentation [3, Sec. 2].
preserve [see also: keep, maintain] The properties of A are preserved in the passage from V to
C(V ). an orientation preserving homeomorphism These are precisely the linear functionals
which also preserve multiplication, i.e., f(ab) = f(a)f(b).
presume [see also: assume] We presume a basic knowledge of large cardinals and forcing.
prevent [see also: preclude, rule out, obstacle, obstruction] The local homeomorphism property
of F does not, however, prevent it from self-intersections. The presence here of the direct
summand H is simply to prevent A from having disconnected spectrum.
previous [see also: foregoing, above] in the previous chapter [= in the immediately preceding
one] in a previous chapter [= in an earlier chapter] This is in agreement with our previous
notation. In this chapter we shall depart from the previous notation and use the letter m not
for Lebesgue measure, but for Lebesgue measure divided by (2)
1/2
. Analysis of the proofs of
these previous results shows that..... We now apply the previous observation to estimate F.
The primary advance is to weaken the assumption that H is C
2
, used by previous authors, to
the natural condition that H is C
1
.
previously [see also: above, before] Here Ff is the previously dened Fourier transform of f.
Our result extends previously known results for nilpotent groups. In our next theorem, we
state a characterization of..... which does not seem to have been noticed previously.
price [see also: cost, expense] We can make g Lipschitz at the price of weakening condition (i).
primarily [see also: mainly, mostly, largely, part] The dierence between these maps is pri-
marily in their kneading sequences.
primary [see also: crucial, key, main, major, principal, basic, elementary] a matter of primary
importance the primary purpose
prime .....where the prime means that..... ....., the prime denoting the omission of the zero
term. the set of all primes not exceeding k
principal [see also: crucial, key, main, primary] Our principal tools in the proof are..... The
proof is similar to the proof of Theorem 4, with two principal modications.
prior Indeed, to our knowledge, cardinality restrictions on Berg spaces have received no prior
attention outside the metric context.
priority With no parentheses, and take priority over the group operation.
probability Dene a
k
to be the probability that exactly k out of the 2n values X
i
exceed T,
conditional on X
0
> T.
probably There are probably many injective E-modules that are p-torsion for all n. The
following proposition is probably well known, but we do not have a reference.
MATHEMATICAL ENGLISH USAGE 111
problem [see also: complication, trouble, issue, subject, matter, topic, point] The main prob-
lems that we address are..... The problem one runs into, however, is that f need not be.....
The problem is to move all the discs to the third peg by moving only one at a time. However,
even for k = 1 the problem proved quite dicult. However, we immediately encounter the
problem of nonregularity of the data. In the nal section of the paper, we list some open prob-
lems. We close this paper by oering some questions and problems for further research. To
overcome this problem, we revise our denition of a branch. We can now pose a problem whose
solution will aord an illustration of how (5) can be used. The factor Gf poses no problem
because..... Pointwise convergence presents a more delicate problem.
procedure [see also: approach, method, technique, way, scheme, line] This procedure can
be extended to take care of any number of terms. This procedure, once implemented, can
thereafter be applied with great eectiveness. Our procedure will be to nd..... Repeating
this procedure enough times gives the desired triangulation. Hochberg pointed to the need for
procedures that are more powerful than classical comparison methods.
proceed [see also: act, operate, continue, pursue, move, turn] The proof proceeds along the
same lines as the proof of Theorem 4. The standard proofs proceed via the Cauchy formula.
We now proceed to matrix rings. In Section 3 we proceed with the study of..... [= start or
continue the study] We shall proceed without making explicit distinctions between the two
types of convergence. Proceeding further in this direction, we obtain the following corollary.
Proceeding as in the proof of....., let..... Before proceeding we record an inequality for the size
of an admissible X.
process [see also: course] Suppose that the process continues indenitely. Continuing this
process, we get..... This process can be repeated until we obtain the promised triangulation.
It is impossible to predict the eventual outcome of the process. In the process of replacing
each A by a smaller set, we cannot make B so small that m(B) < 1.
produce [see also: give, furnish, provide, present, yield, construct, create] By induction, this
process produces a sequence (x
n
) such that..... This result shows that the mere existence of
a nontrivial automorphism j of M produces the cut I(j) of M that satises (2). The main
new feature is the use of the face ring to produce lower bounds for the number of vertices.
Roughly speaking, we shall produce a synthesis of index theory with Fourier analysis. This
metric produces the usual topology of X. Here, of course, the set A produced is rather thin
and certainly nowhere near the densities we are looking for.
product We can factor g into a product of irreducible elements. Note that the apparently
[= seemingly] innite product in the denominator is in fact nite. Then P is the product of
several integer factors of about x
n
in size. We need to check that F-derivatives behave in the
way we expect with regard to sums, scalar multiples and products. A dierent notation is used
because the usual tensor product symbol is reserved for the tensor product of A-bimodules.
Then B(E) is a unital Banach algebra with product the composition of operators.
professor In 1935 he was promoted to the rank of associate professor at Cornell. He is currently
Professor of Mathematics at Texas State University. He spent six months as a visiting professor
at Brown University.
profound If nothing else, I hope to convince my readers that Segals theorem deserves recognition
as a profound contribution to Gaussian analysis.
112 JERZY TRZECIAK
profusion [see also: abundance, abound, numerous, plentiful, variety] We now show that BL-
algebras exist in great profusion.
progress [see also: improvement] Recently, progress has been made in this case by Barnes
(1999). work in progress considerable encouraging/remarkable/signicant/substantial/
impressive/further) progress
project 1 This is part of a larger project to study the Galois groups of periodic points of arbitrary
polynomial maps.
2 The measure m must project onto m
i
under P.
promise We now full the promise made at the start of the proof by handling the case of G = S
n
.
This process can be repeated until we obtain the promised triangulation.
prompt [see also: motivate] This choice was prompted by substantial numerical evidence.
This observation prompted the author to look for a more constructive solution.
proof [see also: evidence, verication, check] Here is a simple direct proof. The others being
obvious, only (iv) needs proof. The proof will only be indicated briey. We can assume that p
is as close to q as is necessary for the following proof to work. The proof follows very closely the
proof of (2), except for the appearance of the factor x
2
. The proof proper [= The actual proof]
will consist of establishing the following statements in sequence. The standard proofs proceed
via the Cauchy formula. An ingenious alternative proof, shorter but still complicated, can be
found in [MR]. Kim announces that (by a tedious proof) the upper bound can be reduced to
10. The following has an almost identical proof to that of Lemma 2. A close inspection of the
proof reveals that..... This nishes completes) the proof. The method of proof carries over
to domains satisfying..... This sort of proof will recur frequently in what follows. We end this
section by stating without proof an analogue of..... It seems reasonable to expect that....., but
we have no proof of this. a laborious complicated/routine/straightforward) proof
proper The proof proper [= The actual proof] will consist of establishing the following statements
in sequence.
properly [see also: strictly] Let K and L be quasivarieties such that K is properly included
in L.
property [see also: characteristic, feature] Let f be a map with f[M having the Mittag-Leer
property. Then F has the property that..... the space of all functions with the property
that..... Now F has the additional property of being convex. We have to show that the
property of there being x and y such that x < y uniquely determines P up to isomorphism.
The operators A
n
have still better smoothness properties. Consequently, F has the
2
property.
[= F has property
2
] Among all X with xed L
2
norm, the extremal properties are achieved
by multiples of U. However, not every ring enjoys the stronger property of being bounded.
On the other hand, as yet, we have not taken advantage of the basic property enjoyed by S:
it is a simplex. Certain other classes share this property. This property is characteristic of
holomorphic functions with..... The structure of a Banach algebra is frequently reected in
the growth properties of its analytic semigroups. It has some basic properties in common with
another most important class of functions, namely, the continuous ones. The space X does
not have fails to have) the Radon-Nikodym property. Only for very heavy-tailed data is this
property violated. [Note the inversion.]
MATHEMATICAL ENGLISH USAGE 113
proportion [see also: fraction, ratio] All sides were increased by the same proportion. The
proportion of M with h
M
= 1 is between 70% and 80%. We show that nevertheless a positive
proportion of the polynomials B
n
(x) satisfy Eisensteins criterion. The proportion of men to
women in the population has changed in recent years. The chart shows government spending
expressed as a proportion of national income.
propose [see also: suggest] He was the rst to propose a complete theory of triple intersections.
A model for analysing rank data obtained from several observers is proposed. Unfortunately,
the proposed model does not satisfy condition (5).
proposition It is this proposition that we believe to be false in Morava E-theory.
prove [see also: show, establish, demonstrate, evidence, substantiate, validate] The theorem to
be proved is the following. [= which will be proved] The uniqueness of f is easily proved,
since..... If n = 1, there is nothing to prove. However, this cannot be proved of the cardinal
function d(X). Much to our surprise, even for k = 1 the problem proved quite dicult. This
result will prove extremely useful in Section 2. The conjecture will be disproved by exhibiting.....
provide [see also: give, aord, furnish, oer, produce, supply, yield] Our last example was
kindly provided by B. Johnson. This itself does not produce a solution of (1), but an additional
hypothesis such as the Palais-Smale condition does provide such a solution. This provides an
eective means for computing the index. In addition to illustrating how our formulas work
in practice, it provides a counterexample to Browns conjecture. The method falls short of
providing an explicit formula for the index. This will provide us with a way of getting some
information about..... without relying on general theory.
provided [= providing; see also: if] This equation has a solution in integers provided that
N > 7. The resulting metric space consists precisely of the Lebesgue integrable functions,
provided we identify any two that are equal almost everywhere.
provisional [see also: temporary, ad hoc] We make the following provisional denition, which
is neither general nor particularly elegant, but is convenient for the induction which is to follow.
publish Only a few of those results have been published before.
purpose [see also: aim, end, object, objective, intention] Let us now take a quick look at the
class N, with the purpose of determining how much of Theorems 1 and 2 is true here. For
this purpose, we rst dene..... For our purposes here, the best way is to base the proof on
the following theorem, a derivation of which can be found in [P]. The purpose of this paper
is to expand substantially the class of maps for which the index can be computed. A detailed
exposition, more suited to the purposes of the present article, is given in [9]. The motivation
for the results of this section is the following result of John (paraphrased slightly to suit our
purposes).
pursue [see also: proceed, go on, continue] We shall pursue our investigation of conservation
laws in Section 5. This question is pursued further in Section 3. While we suspect that the
methods of the present section can be brought to bear on this question also, we shall not pursue
this analysis in detail here.
push This assumption enables us to push through the same arguments.
114 JERZY TRZECIAK
put [see also: insert, plug, set] Put a subset U of j(X) in T if its inverse image under j is an
open subset of X. Put a taxicab metric on S
k
. The map F can be put into this form by
setting..... Put this way, the question is not precise enough. This puts a completely dierent
perspective on Foxs results. This puts us in a position to apply Lemma 2 to deduce that.....
We put o discussing this problem to Section 5. Putting these results together, we obtain the
following general statement.....
Q
quality To assess the quality of this lower bound, we consider the following special case.
quantitative Theorem 7 imposes a quantitative restriction on the location of the zeros of.....
quantity Theorem 1 relates the quantity a(n) to the quantity b(n). A natural question to
ask is how the quantities A(S, T) and B(S, T) vary as S and T change. The quantities F
and G dier by an arbitrarily small amount. Generally we add a tilde to distinguish between
quantities associated with

G and those associated with G.
question [see also: problem, issue, matter, subject, topic, point] An obvious question to ask is
whether Theorem 1 continues to hold for..... A natural question is how sharp the bounds given
in Theorem 6 are. The question we shall be concerned with is whether or not f is..... The
question naturally arises whether this representation is unique. Here the interesting questions
are not about individual examples, but about the asymptotic behaviour of the set of examples
as one or another of the invariants (such as the genus) goes to innity. The question of whether
B is ever strictly larger than A remains open. [Or: The question whether] Incidentally, the
question of whether K(E) is amenable for specic Banach spaces E seems to have received
almost no attention in the literature. The question as to how often we should expect the
class number to be divisible by p is also of some interest. Put this way, the question is not
precise enough. This question was answered negatively in [5]. The two questions listed below
remain unanswered. This brings about the natural question of whether or not there is any
topology on the set of all possible itineraries. The lemma raises an interesting question:.....
This suggests a question: under what conditions is it true that..... ? We close this paper by
oering some questions and problems for further research. It is generally a highly nontrivial
question whether..... When A is commutative, the answer to both questions is yes. The
continuum in question is then called arc-like.
quickly With c = 4, it is quickly seen that R(q) decreases.
quite [see also: fairly, somewhat, rather] There is quite an extensive literature concerning
resonance problems, beginning with the work of Lazer. The two characteristics are connected,
but the relationship is quite a complex one. There are quite a number of cases, but they can
be described reasonably systematically. This is essentially implicit in [BC] but we cannot quite
quote the result we need. For binary strings, the algorithm does not do quite as well.
quote [= repeat exact words; see also: cite] We begin by describing the class of functions f
considered, which includes the special cases quoted above. We quote for future reference another
result of Fox: there exists..... In Section 2 the reader will be reminded of some important
properties of Bernoulli numbers, and some auxiliary results will be quoted or derived. This is
essentially implicit in [BC] but we cannot quite quote the result we need.
MATHEMATICAL ENGLISH USAGE 115
R
radius [pl. radii] a ball of radius r about the origin
raise [see also: increase] This raises the following question. [Not: rises] In [K] we raised
the question of recovering X from X
M
. Raise both sides to the power p to obtain..... The
maximum velocity was raised to V .
random The random variable X has the Poisson distribution with mean v. In this and the
other theorems of this section, the X
n
are any independent random variables with a common
distribution. To calculate (2), it helps to visualize the S
n
as the successive positions in a random
walk. The proof shows that if the points are drawn at random from the uniform distribution,
most choices satisfy the required bound.
range 1 [see also: scope, extent, reach] Theorem 2 still holds for A(x) provided that k is
restricted to the range [0, 1]. for p in the range 1 < p < for k in the indicated range
We also show the existence of E-transformations exhibiting nearly the full range of behaviours
possible for scaling transformations.
2 [see also: vary] .....where h ranges over the set of elements having..... The elements of G,
numbering 122 in all, range from 9 to 2000.
rank a rank one operator an operator of nite rank = a nite rank operator
rapidly [see also: fast] In the study of innite series

a
n
it is of signicance whether the a
n
approach zero rapidly. a rapidly decreasing function
rare [see also: unusual] Conditions relating to bounds on the eigenvalues appear to be rare
in the literature. Indeed, as n increases, it becomes increasingly rare for a manifold to be a
hyperplane section of another projective manifold.
rarely While C
p
can be a Baire space, it rarely has the stronger completeness properties men-
tioned above. Berg spaces have been rarely considered outside the metric context.
rarity We also need the following technical lemma, which asserts the rarity of numbers with an
inordinately large number of prime factors.
rate These estimates only require that f have a a certain polynomial rate of decay at innity.
We give a fairly simple description of a wide class of averaging operators for which this rate
of growth can be seen to be necessary. Moreover, one has estimates on the rate at which this
convergence is taking place. the growth rate of V
n
as n
rather [see also: fairly, somewhat, quite, instead] However, F is only nonnegative rather than
strictly positive, as one may have expected. In fact, we shall prove our result under the weaker
hypothesis that W is weakly bounded, rather than just bounded, on an innite subset of G.
The denition is stated in terms of local martingales, rather than martingales, for the technical
reason that the former are easier to characterize in applications. It seems that the relations
between these concepts emerge most clearly when the setting is quite abstract, and this (rather
than a desire for mere generality) motivates our approach to the subject. Rather than working
directly with V (s), we shall instead consider the following two general integrals:..... [Or: Rather
than work] Rather than discuss this in full generality, let us look at a particular situation
of this kind. The proof is rather cumbersome. This may appear rather wasteful, especially
when n is close to m, but these terms only give a small contribution to our sum. For explicit
116 JERZY TRZECIAK
solutions, it may be necessary to have rather precise information about the amplitude . We
rst prove the (rather simpler) Theorem 7, by eecting a quite general reduction of the problem
to the study of certain isotropy factors. A further complication arises from BP, which works
rather dierently from the other labels.
ratio [see also: proportion] The female/male ratio in the sample was 19 to 1. the ratio of P
to Q the ratio between Haar measures on G and H in the ratio of approximately 3 : 1
reach 1 We would like to know..... but that is beyond our reach at this point. It seems that
the solution of Problem 1 is still out of reach.
2 [see also: achieve, attain, arrive at] The point A can be reached from B by moving along an
edge of G. For example, F reaches a relative maximum of 5.2 at about x = 2.1. The iterates
eventually reach the value 1.
read Equation (2) now reads Ax =..... The author thanks H. Miller for a careful reading of an
earlier draft. The string N (read from right to left) starts with..... Then G is simply g with
its periodic string read backwards. This can be read o from (8). We do this in the rst
section, which the reader may skip on a rst reading.
readability We have made modications in the interest of readability.
readable A very readable account of the theory has been given by Zagier [3].
reader We leave the details to the reader. We leave it to the reader to verify that..... [Note that
the it is necessary here.] For more details we refer the reader to [4]. If nothing else, I hope
to convince my readers that Segals theorem deserves recognition as a profound contribution to
Gaussian analysis. It may be worth reminding the reader that..... The reader is cautioned
that our notation is in conict with that of [3]. The reader is assumed to be familiar with
elementary K-theory. The interested reader is referred to [4] for further information. [Note the
double r in referred.] At this point, the reader is urged to review the denitions of..... The
reader might want to compare this remark with [2, Cor. 3]. For the convenience of the reader,
we repeat the main points. [Or: For the readers convenience]
readily [= without diculty] We have shown that....., whence it is readily inferred that..... One
of the appealing aspects of the spectral set is that it readily lends itself to explicit computation.
ready [see also: position] We are now ready to proceed to the nal stage of our construction.
realize [see also: know, recognize, understand, grasp, aware] The theorem gains in interest if
we realize that..... This has deeper signicance than one might rst realize. The geodesics (8)
are the only ones that realize the distance between their endpoints. In the remainder of this
section, we study some properties of K, with the eventual aim (not realized yet) of describing
K directly using G.
really [see also: actually, fact] We did not really have to use the existence of T. Of course, it
is tacitly understood that it is this measure that is really under discussion. Each x here really
designates the pair (x, Ax).
rearrange Rearranging terms we obtain the inequality.....
rearrangement The obvious rearrangement reveals the right side to be identical with (8).
MATHEMATICAL ENGLISH USAGE 117
reason 1 [see also: explain, justify, ground] This argument is invalid for several reasons. It
is for this reason that his argument is incorrect. The denition is stated in terms of local
martingales, rather than martingales, for the technical reason that the former are easier to
characterize in applications. This is the reason for calling f the derivative of g. One reason
for using this rather elaborate model is that it permits a simple and concrete denition of the
realization. A drawback to Polya trees, and perhaps the main reason why they have not seen
much application within the Bayesian nonparametric literature, is that an arbitrary partition
tree needs to be specied. The reason that this is signicant is that..... Indeed, there is reason
to suspect that diculties could increase with increasing n. There is no reason to expect this
to be an inverse map on K, but we do have the following.
2 [see also: argue, proceed] To see this, we reason as follows. The simplest way is to reason
by induction on n.
reasonable [see also: mild, moderate] It is therefore reasonable that the behaviour of p should
in some rough sense approximate the behaviour of q. Thus it is reasonable to attempt, using
this homeomorphism, to gain an understanding of the structure of M. It seems reasonable to
expect that....., but we have no proof of this. If the boundary is never hit then x
t
is a Feller
process under reasonable continuity assumptions.
reasonably [see also: fairly] Our asymptotic results compare reasonably well with the numerical
results reported in [8]. There are quite a number of cases, but they can be described reasonably
systematically.
reasoning [see also: argument] By the same kind of reasoning it suces to consider..... Our
method of proof will be an adaptation of the reasoning used on pp. 7172 of [3]. The same
line of reasoning applies in the continuous time setting. We could have adopted an approach to
proving Theorem 2 along a line of reasoning which bears greater resemblance to the treatment
of the analogous result in Section 1. To recover Wieners famous result that Brownian paths
are continuous, one needs to use more sophisticated reasoning.
reassemble [= put together again] The diagram of L+M is obtained by taking the rows of the
diagrams of L and M and reassembling them in order of decreasing length.
recall [see also: remind, remember] We recall what this means. [Not: We remind] Recall
the denition of T from Section 3.
receive [see also: get, obtain] He received his master and Ph.D. degrees from the University
of Texas. Incidentally, the question of whether K(E) is amenable for specic Banach spaces
E seems to have received almost no attention in the literature. Indeed, to our knowledge,
cardinality restrictions on Berg spaces have received no prior attention outside the metric context.
recent For a recent account we refer to [4]. However, the connection with Gromovs work
has been obscured in recent years by an emphasis (in the algebraic topology literature) on
conguration spaces.
recently This subject has recently been extensively studied. There has recently been increasing
interest in the theory of..... The authors interest in this problem was recently rekindled by a
conversation with David Lees. This method is recently less and less used. There are, however,
a few important papers of which we were unaware until fairly recently. Very recently, Heck
introduced a general approach that unies and extends all these results.
118 JERZY TRZECIAK
recognition The name of Harald Bohr is attached to bG in recognition of his work on almost
periodic functions. If nothing else, I hope to convince my readers that Segals theorem deserves
recognition as a profound contribution to Gaussian analysis.
recognize [see also: identify, know, nd, establish, realize, grasp, understand, admit, acknowl-
edge] It is important to recognize that there is a signicant subclass of vector elds for which
the ambiguity alluded to above disappears. It is obvious that the above theorem supplies an
algorithm to eectively recognize whether SP is in A. Because N. Wiener is recognized as
the rst to have constructed such a measure, the measure is often called the Wiener measure.
a method for recognizing pure injective modules
recommend [see also: suggest] The author thanks the referee for recommending various im-
provements in exposition.
record Before proceeding we record an inequality for the size of an admissible X. For later
use, we record the following formulas:..... For future reference, we record this in the following
corollary. Standard Banach space notation is used throughout. For clarity, however, we record
the notation that is used most heavily.
recourse [see also: appeal, invoke, refer] Let us now prove directly (without recourse to [5])
that..... This case can in fact be treated without recourse to the methods of that paper.
recover [see also: retrieve] If we know a covering space E of X then not only do we know
that....., but we can also recover X (up to homeomorphism) as E/G. Thus F can be recovered
from X
k
F by k-fold integration. Replacing f by log f, we recover the theorem of [6]. This
is nearly the same as formula (6) of [7], which we can recover by multiplying (2.3) by F(s).
To recover Wieners famous result that Brownian paths are continuous, one needs to use more
sophisticated reasoning.
recur This sort of proof will recur frequently in what follows.
rederive The inequality (2.4) is essentially contained in [LS] but will be rederived in Corollary 5
below.
reduce [see also: decrease, diminish, lower, cut down] We claim that, by setting w to zero on
this interval, the value of F(w) is reduced. Kim announces that (by a tedious proof) the upper
bound can be reduced to 10. The length of F is thus reduced by half. By induction, we are
reduced to proving the following lemma. The problem now reduces to establishing that.....
We rst prove a reduced form of the theorem.
reduction A small percentage reduction in the cost of materials resulted in a signicant increase
in prot. We rst prove the (rather simpler) Theorem 7, by eecting a quite general reduction
of the problem to the study of certain isotropy factors.
redundant [see also: superuous, unnecessary, drop, omit] Note that no boundedness assump-
tion is made in this denition; in fact, this would be redundant as shown by Theorem 3 below.
refer [see also: mention, allude, touch, appeal, invoke, recourse] For more details we refer the
reader to [4]. We refer the reader to the body of the paper for details. For a comprehensive
treatment and for references to the extensive literature on the subject one may refer to the
book [M] by Markov. The interested reader is referred to [4] for further information. [Note
the double r in referred.] In the physical context already referred to, K is the density of.....
MATHEMATICAL ENGLISH USAGE 119
This theorem will hereafter be referred to as the minimum principle. We refer to these as
homogeneous Sobolev spaces. We shall also refer to a point as backward nonsingular, with the
obvious analogous meaning. In the following, all topological notions refer to the weak topology
of Y . To be precise, refer to the notation in Theorem 1 and dene.....
referee 1 The author thanks the referee for his helpful suggestions concerning the presentation
of this paper. The author thanks the referee for recommending various improvements in expo-
sition. We are grateful to the referee for a number of helpful suggestions for improvement in
the article. The referee deserves thanks for careful reading and many useful comments. At
the suggestion of the referee, we consider some simple cases.
2 While the rst version was being refereed, I found that Zhang [2] had given a similar treatment
of E
n
(X).
reference Let us note (for later reference) that..... We quote for future reference another result of
Fox: there exists..... For these considerations it will be convenient to state beforehand, for easy
reference, the following variant of..... We have thus proved the theorem without any reference
to integration. If G is clear from context, then we suppress reference to it in the notation. The
best general reference An excellent recent reference) for..... is [5]. The following proposition
is probably well known, but we do not have a reference. At the time of writing [5], I was not
aware of this reference. Where it is possible, we outline the proofs so that the reader will not
have to hunt for obscure references. For a comprehensive treatment and for references to the
extensive literature on the subject one may refer to the book [M] by Markov.
rene [see also: improve, sharpen, strengthen, progress, sophisticated] In that case, we may
rene Lemma 3 to the following more precise statement. Slightly rening a result of [8],
Davenport proved..... a succession of more and more rened discrete models
renement [see also: improvement, strengthen] For later use, we prove a mild renement of
this latter characterization.
reect [see also: represent, reveal, show, indicate] The structure of a Banach algebra is fre-
quently reected in the growth properties of its analytic semigroups. Strictly speaking, we
should write something like a(l, m, n) to reect the dependence; we shall rely upon context in-
stead. In particular, integral curves evolve continuously, and we should seek to represent them
using a measure which reects this continuity in some way. If s
0
lies below R
2
, then we can
reect about the real axis and appeal to the case just considered.
reection A little reection on the denitions makes it clear that.....
reformulate [see also: rephrase, restate, rewrite] This can be easily reformulated in purely
geometric terms.
reformulation A similar reformulation can be made for.....
refute [= prove that sth is incorrect] However, this argument can be easily refuted by showing
that.....
regard 1 [see also: respect] We need to check that F-derivatives behave in the way we expect
with regard to sums, scalar multiples and products. See Section 3 in this regard.
2 [see also: consider, concern, view] We can regard (8) as an equation for . Regarded as the
120 JERZY TRZECIAK
intersection of two quadrics, E represents..... We also discuss some intriguing open questions
regarding triply periodic surfaces. Regarding (8) [= Concerning (8)], we have.....
regardless [of sth; see also: irrespective] This works regardless of whether B is true or false.
Their result gives no information when k is large, whereas (5) is signicant regardless of the size
of k.
relabel [see also: rename] Next we relabel the collection A
n
, B
n
as C
n
. Then we can nd
a subsequence (not relabelled) such that a
n
< 1 for all n.
relate [see also: concern, connect, link, tie, associate, pertain, relevant] We describe how the
notion of positivity relates to the other properties. Conditions relating to bounds on the
eigenvalues appear to be rare in the literature. Theorem 1 relates the quantity a(n) to the
quantity b(n). [Not: with the quantity b(n)] There are numerous results in the literature
relating spectral conditions to invertibility of f. The following proposition relates the two
denitions. There is a related result concerning primitivity. We rst replace a(n) by the
related and simpler function b(n), where..... Our results are closely related to those of Strang [5].
The above-mentioned measure is of course intimately related to the geometry of the real line.
How are these two optimality notions related? In general, they are not.
relation [see also: connection, link, relationship] But there is a much more important relation
between equivalent regions:..... By carefully examining the relations between the quantities U
i
,
we see that..... There is a fourth notion of phantom map which bears the same relation to the
third denition as the rst does to the second. What relation exists between f and g?
relationship [see also: connection, link, relation] On the way we analyze the relationship be-
tween..... The two characteristics are connected, but the relationship is quite a complex one.
The aim of this article is to study the relationship between the size of A, as measured by its
diameter, and the extent to which A fails to be convex.
relative [see also: respect, compare] Then P is said to be elliptic relative to the action of G
if..... [Not: relatively to] Then G is a subgroup of R (relative to addition). The pull-back
of F is homotopic to G relative to the end-points. the complement of A relative to B Here
A is small relative to B. [= in comparison with B]
relax [see also: weaken] The idea is to relax the constraint of being a weight function in Theo-
rem 3. The assumption that the test statistics are identically distributed can be relaxed without
much diculty.
relevance [see also: role, importance, signicance] What is the relevance of this example to
Fatous lemma? To understand the relevance of locality, note that it implies that.....
relevant [to sth; see also: pertinent, connect, concern, pertain] We now turn to a brief discussion
of another concept which is relevant to Johns theorem. This comment is relevant in proving
Theorem 1. For a list of relevant references, see [2]. [= references connected with the subject
being considered] For relevant background material concerning random walks, see [2].
reliance [see also: dependence] The proof is self-contained, without reliance on a computer-
algebra system.
rely [see also: base, depend] The proof is not direct, but relies on the results of [2]. We
underline that the aforementioned results in [1] all rely on the conformality of the underlying
MATHEMATICAL ENGLISH USAGE 121
construction. This will provide us with a way of getting some information about M without
relying on general theory. Strictly speaking, we should write something like a(l, m, n) to reect
the dependence; we shall rely upon context instead.
remain [see also: stay, keep, continue, persist] Conjecture 2 of [KH], to the eect that [= mean-
ing that] there is no relation P with E(P) = 1, still remains open. The case where p > 1
remains unresolved. For k = 2 the count remains as is. The two probabilities remain essen-
tially what they were before. The situations with domains other than sectors remain to be
investigated. It remains to exclude the case where..... To prove (8), it only remains to verify
that..... Thus, all that remains is to repeat the construction for f in place of g. There remains
the second question. [But: It remains to consider the second question.] It is possible that the
methods of this paper could be used to....., but there remain considerable obstacles to overcome.
Let S
i
be the rst of the remaining S
j
. Half of the sets of the family R miss i and half the
remaining miss j. The remaining possibility is that v is labelled 2.
remainder [see also: rest] The remainder of our work breaks into ve steps. In the remainder
of this section we shall be trying to answer the question:..... The map f is xed for the remainder
of the proof.
remark 1 [see also: observation] However, the ambiguity alluded to in Remark 3 disappears
when talking about an ane eld. The reader might want to compare this remark with [2,
Cor. 3]. First, some remarks concerning the term closure are in order. This is no coincidence,
in the light of the remarks preceding Denition 2. Following these preliminary remarks, we
now state.....
2 [see also: note, observe, comment] It is perhaps worth remarking that..... We remark at the
outset that this formula makes sense, because..... However, this argument is fallacious, because
as remarked after Lemma 3,.....
remarkable [see also: unusual] Their remarkable achievement seemed to validate Johns claim.
However, it soon turned out that..... The remarkable feature of this theorem is that.....
Theorem 3 is remarkable in that considerably fewer conditions than in the previous theorems
ensure universality.
remarkably In this paper we apply combinatorial group theory (sometimes in a remarkably
elementary way) to prove a number of results in this direction.
remedy In this case we need to take into account the diculty with tting a ball into a sector.
The obvious remedy is to replace the ball by a suitable disc.
remember [see also: recall, reminiscent, mind] To understand why, let us remember that.....
Theorem 3 may be interpreted as saying that A = B, but it must then be remembered that.....
remind [sb that sth; sb of sth; see also: recall] It may be worth reminding the reader that.....
[Not: reminding that] The purpose of this section is to remind the reader of some of the
results on the structure of..... In Section 2 the reader will be reminded of some important
properties of Bernoulli numbers, and some auxiliary results will be quoted or derived.
reminiscent [of sth; see also: similar] It has properties reminiscent of partition functions.
Away from critical points, the action of G is reminiscent of the action of a cyclic group of
order d.
122 JERZY TRZECIAK
removal Likewise, if A does not span C(I), removal of any of its elements will diminish the span.
remove [see also: delete, drop, omit, dispense with] This theorem removes the restriction
to convex regions which was imposed in Theorem 8. The theorem implies that some nite
subcollection of the f
i
can be removed without altering the span. Let D be the plane with
three disjoint discs removed.
rename [see also: relabel] Then we can nd a subsequence (not renamed) such that a
n
< 1 for
all n.
renewal This article features results in both spectral theory and operator ergodic theory made
possible by a recent renewal of interest in the consequences of Jamess inequalities.
repeat Thus, all that remains is to repeat the construction for f in place of g. Repeating this
procedure enough times gives the desired triangulation. Note that some of the a
n
may be
repeated, in which case B has multiple zeros at those points. This process can be repeated
until we obtain the promised triangulation. By repeated squaring to eliminate the radicals in
this equality, we obtain.....
repeatedly [see also: frequently, often] This property will be used repeatedly hereafter.
repetition To avoid undue repetition in the statements of our theorems, we adopt the following
convention. The proof is essentially a repetition of the arguments used to prove.....
rephrase [see also: reformulate, restate, rewrite] As usual, we can rephrase the above result
as a uniqueness theorem. Rephrased in the language of [HT], Proposition 2 says that.....
A rephrasing of the denition is that.....
replace [see also: change, place, instead] We now construct f as in the proof of Theorem 5, with
V replaced by W. A similar result holds with compact replacing convex. The statement
of Theorem 5 remains valid if we replace f is compact by the norm of f is bounded. We
must have Lf = 0, for otherwise we can replace f by f Lf.
replacement Then F satises the following replacement for condition (e). Replacement of z
by 1/z transforms (4) into (5). We produce an evolution equation which diers from (2.3) only
in the replacement of the F
2
term by F
3
.
report Our asymptotic results compare reasonably well with the numerical results reported in [8].
represent [see also: present, reect, picture, account] We can represent W by the integral.....
We represent A as a quotient space of X by sending the point..... to..... [Not: We present]
It turns out that A, B and C all belong to the same class, which we represent by the symbol P
2
.
In particular, integral curves evolve continuously, and we should seek to represent them using
a measure which reects this continuity in some way. We shall use the symbol to represent
intersection. We use upper case letters to represent inverses of generators. This accords
with the intuition that as we pass down the coding tree, we nd out more and more detailed
information about the ordering actually represented. We shall then show that this f can be
represented in the form (5).
representative a complete set of representatives of the isomorphism classes of A-modules
require [see also: call for, entail, demand, want, desire, necessitate, force, involve, requisite]
Then f = 1, as required claimed/desired). In addition to f being convex, we require that F
MATHEMATICAL ENGLISH USAGE 123
be holomorphic. [Note the subjunctive be.] This is equivalent to requiring that..... If (ii) is
required for nite unions only, then M is called an algebra of sets. When clarity requires it,
we shall write I
A
(f) instead of I(f). [Note that the it is necessary here.] The above denition
was rst given in [D]; care is required because this term has been used in a slightly dierent
sense elsewhere. Many of the results that follow only require W to satisfy (8). The algebraic
properties (i) and (ii) required of j are evidently true. Lemma 3 does not require D to be
prime. The only point that requires care is the verication of..... The projection technique
requires the introduction of an appropriate homomorphism. We require a method of dividing
z into subwords z
i
for which we know the structure of f(z
i
). This requires us to give M an
R-module structure. We shall show that V is the required open cover.
requirement [see also: condition, demand, stipulation] The conclusion of Theorem 3 be-
comes false if this requirement is omitted. We next show how the continuity requirement
on f in Theorem 2 can be weakened. In less precise language, the requirement is that the
two angles are the same in size and in orientation. The remaining requirements of Deni-
tion 3 are also easily seen to hold in L
1
. The other requirements for a type F map are
also met. Thus W satises two of the four requirements for membership in Z. the main
key/essential/detailed/precise/reasonable/minimum/further/specic) requirement
requisite [= necessary for a particular purpose; see also: require] In order to prove the opposite
inequality, we rst construct the requisite number of balanced spheres of type A.
research [see also: investigation, study] Research partially supported by NSF Grant No. 23456.
This research was initiated when the rst author was visiting the University of Alberta in
the summer of 2008; the nancial support and kind hospitality are gratefully acknowledged.
A survey of the research on f
n
(x, y) up to 1970 (most of it dealing with the case n = 1) was given
in [3]. This is an interesting area for future research. We close this paper by oering some
questions and problems for further research. This work continues research begun in [5]. In the
present work we continue this line of research. detailed in-depth/extensive/pioneering/ground-
breaking/thorough) research
resemblance [see also: similarity] We could have adopted an approach to proving Theorem
2 along a line of reasoning which bears greater resemblance to the treatment of the analogous
result in Section 1. All three cases bear a striking resemblance.
resemble [see also: like, similar] Littlewood proved that certain subharmonic functions on the
unit disc resemble bounded analytic functions in having radial limits almost everywhere. While
topological measures resemble Borel measures, they in general need not be subadditive.
reserve The letter will be reserved for characteristic functions throughout this book. A dier-
ent notation is used because the usual tensor product symbol is reserved for the tensor product
of A-bimodules.
resistant On the other hand, the second case appears to be resistant to the methods we use to
prove Lemma 2.
resolve [see also: solve] However, to our knowledge this is not fully resolved. This also resolves
the ambiguity introduced earlier in choosing an order of the lifts of U. The case where p > 1
remains unresolved.
124 JERZY TRZECIAK
respect [see also: regard, relative, compare] The set S is a semigroup with respect to coor-
dinatewise addition. Thus C behaves covariantly with respect to maps of both X and G.
Suppose A is maximal with respect to having connected preimage. We denote the complement
of A by A
c
whenever it is clear from the context with respect to which larger set the comple-
ment is taken. Now suppose that F(n) = x and that n is maximal in this respect. The
prime 2 is anomalous in this respect, in that the only edge from 2 passes through 3. It is in
all respects similar to matrix multiplication. In the course of writing this paper we learned
that P. Fox has simultaneously obtained results similar to ours in certain respects. Keller [2,
Theorem 5] obtains a duality theorem that is stronger than Theorem 2 in a number of respects,
but the proof is much more dicult. Identical conclusions hold in respect of the condition BN.
[= concerning BN]
respective [= relating separately to the individual objects just mentioned, in the same order;
see also: correspond] Let x
1
, . . . , x
n
be the distinct values assumed by f and A
1
, . . . , A
n
the
respective sets where these values are assumed. Dierentiation of (5) and (6) gives the respective
equations A = B and C = D.
respectively Let x
1
, x
2
be variable points in the intervals (a, b), (c, d) respectively. The sub-
spaces of M() consisting of the discrete and continuous measures are M
d
() and M
c
(),
respectively. We call E
x
and E
y
the x-section and y-section respectively of E. These might
be called respectively the regular and the singular parts of B.
rest 1 [see also: remainder] F = 1 on the rest of A. Each row of A has a single 1 and the
rest of the entries 0. The rest of the proof runs as before. Much of the rest of the paper is
devoted to a general study of.....
2 [see also: base, depend, rely] This rests on a result of Kummer.
restate [see also: reformulate, rephrase, rewrite] Thus the conjecture can be restated as follows.
restraint [see also: constraint, restriction] A computational restraint is the algebraic number
theory involved in nding these ranks, which will typically be more demanding than in our
example of Section 1.
restrict [see also: conne, limit] It is not generally possible to restrict f to the class D. Then
F restricts to a C
0
ow on M. Theorem 2 still holds for A(x) provided that k is restricted
to the range [0, 1]. The supremum in (1) does not change if we restrict ourselves to rational
points in Q. We shall restrict the discussion to plane regions.
restriction [see also: constraint, limitation, restraint] Theorem 7 imposes a quantitative restric-
tion on the location of the zeros of..... However, if B were omitted in (1), the case n = 0 would
imply Nf = 1, an undesirable restriction. Some restrictions must be placed on the behaviour
of f. When A is the order complex of a poset, there are further restrictions on the h-vector
of A. This theorem removes the restriction to convex regions which was imposed in Theorem 8.
All our estimates hold without this restriction. The location of the zeros of a holomorphic
function in a region is subject to no restriction except the obvious one concerning the absence
of limit points in .
restrictive [see also: stringent] The usual denition is more restrictive in that it requires that
a A.
MATHEMATICAL ENGLISH USAGE 125
result 1 [see also: consequence, eect, outcome, achievement] Let n , applying Lemma 5.
The result is..... As a result, B is isomorphic to C. As usual, we can rephrase the above result
as a uniqueness theorem. Slightly rening a result of [8], Davenport proved..... The method
of proof of Theorem B can be adapted to extend the right-to-left direction of Mostowskis result
by showing that..... We now prove a simple fact about semigroups: this is surely a folklore
result. It is hoped that a deeper understanding of these residues will help establish new results
about the distribution of modular symbols. We could have adopted an approach to proving
Theorem 2 along a line of reasoning which bears greater resemblance to the treatment of the
analogous result in Section 1. In the course of writing this paper we learned that P. Fox
has simultaneously obtained results similar to ours in certain respects. There is a related
result concerning primitivity. The results have been encouraging enough to merit further
investigation. We show that one can drop an important hypothesis of the saddle point theorem
without aecting the result. This rests on a result of Kummer.
2 [see also: cause] The resulting formula exhibits u as the Laplace transform of x. [Not: The
obtained formula] We can solve the resulting equations successively for c
1
, ..., c
n
. Their
study resulted in proving the conjecture for k = 1. The demand that each entry be a perfect
square results in nine equations. [Note the subjunctive be.] Less than 1 in p of its points will
result in a quartic with ideal class number p. Lebesgue discovered that a satisfactory theory
of integration results if the sets E
i
are allowed to belong to a larger class of subsets of the line.
[Do not write: It results that if you mean: It follows that.]
retain [see also: keep, maintain, preserve] This extension retains control on..... at the sacrice
of loosing some control on..... The lemma shows that properties of certain subsets of X are
retained by their images under F.
retrieve [see also: recover] We can retrieve H from H

by the formula.....
return [see also: go back, turn back, revert] The algorithm returns 0 as its answer. We shall
return to this central theme in Chapter 7. The usefulness and interest of this correspondence
will of course be enhanced if there is a way of returning from the transforms to the functions,
that is to say, if there is an inversion formula.
reveal [see also: exhibit, show, demonstrate, display, reect] A careful look at the proofs re-
veals that P is measurable. The next two theorems reveal the importance of this concept.
A complete explication of the Fox spaces is warranted, as it will likely reveal further clues to the
dierences between the parabolic and hyperbolic theories. The obvious rearrangement reveals
the right side to be identical with (8).
reverse [see also: converse, opposite] 1 in reverse order the reverse inequality inclusion)
We show that..... by reverse induction on i, starting at i = n and working down to i = 0.
2 None of these implications can be reversed. By reversing the steps above, we see that.....
One is tempted to reverse the order of integrations but that is illegitimate here. with the order
of a and b reversed
revert [see also: return, go back] Reverting to our former notation, we see in particular that.....
review In Section 2 we review [= go over] the separation of variables formula.
revise [see also: alter, change, modify] To overcome this problem, we revise our denition of a
branch. By revising our choice of A if necessary, we may assume that.....
126 JERZY TRZECIAK
revolution the revolution of the earth about round) the sun
rewrite [see also: reformulate, rephrase, restate] Hence (8) can be rewritten in the form.....
right 1 We now prove a lemma which is interesting in its own right. We cannot hope to say
anything about the structure of each isotropy factor as a system in its own right.
2 the right member of (8) = the right side of (8) = the right-hand side of (8) [Or: the right hand
side] the right-hand side expression on the right of (8) = on the right-hand side of (8) The
exact sequence ends on the right with H(X). composition on the right with p Then D lies to
the right of G. Let n
k
be the rst location to the right of the kth decimal place of W that has
a value less than b. Comparisons are done in left-to-right order. The entries in each row are
increasing from left to right. The string N (read from right to left) starts with..... The map
f, which we know to be bounded, is also right-continuous. The method of proof of Theorem
B can be adapted to extend the right-to-left direction of Mostowskis result by showing that.....
the rightmost integral [= the last one on the right]
3 However, as pointed out right after (5),.....
rigorous [see also: precise] To make this argument rigorous, apply (1.12) to.....
rise 1 [see also: lead] Every possible such sequence gives rise to..... [Not: gives raise]
2 [see also: increase] Ination in the rst quarter rose beyond the acceptable level of 5%.
role [see also: part, importance, signicance, relevance, interest] By interchanging exchanging)
the roles of X and Y , it follows that..... Here f takes over the role of the time parameter. It
turns out that these properties play no role in the proof. In closing this section we take up a
result which will play a pivotal role in the characterization of..... Its role is to rule out having
two or more consecutive P-moves (on the grounds that they can be performed in one go). a key
principal/signicant/important) role
root We now multiply (7) by p and take nth roots to obtain.....
rotate This is the lattice packing rotated 45
o
. Let A denote the rectangle B rotated through
/6 in a clockwise direction about the vertex (0, 1).
rotation [see also: turn] As the point z moves around the unit circle, the corresponding J
z
s
are rotations of angle t(z). rotation through /3 through an angle ) a 180
o
rotation
rough [see also: approximate, rudimentary] It is therefore reasonable that the behaviour of p
should in some rough sense approximate the behaviour of q. For most of the proof it suces
to use the rough bound p < 1.
roughly [see also: approximately, about] Roughly speaking, we shall produce a synthesis of
index theory with Fourier analysis. This says (roughly speaking) that the real part of g is.....
Roughly 0.7 comparisons were done for each character.
round 1 [see also: around] It is also tempting to get round get around) this problem by working
with.....
2 To round out the picture presented by Theorem 5, we mention the following consequence of.....
Values computed for the right side of (2) were rounded up in the fourth decimal place.
routine [see also: standard] It is a routine matter a matter of routine) to show that.....
Indeed, it is routine to verify that the index so constructed is independent of the choices made.
MATHEMATICAL ENGLISH USAGE 127
The remainder of the proof is routine. We sketch the rest of the proof, leaving routine details
to the reader.
routinely There are kneading sequences for which the arguments of Section 4 go through rou-
tinely.
rudimentary [= very basic, at the simplest level] The requirement that competitors are sym-
metric will only be used to get some very rudimentary control on the asymptotics of G.
rule 1 However, such situations are the exception rather than the rule. There are few exceptions
to this rule. It follows that any itinerary that obeys these four rules corresponds to a point
in B.
2 [see also: exclude, preclude] The possibility A = B is ruled out in the same way. Its role is
to rule out having two or more consecutive P-moves (on the grounds that they can be performed
in one go).
run The rest of the proof runs as before. Although these proofs run along similar lines, there
are subtle adjustments necessary to t the argument to each new situation. Let A

be A run
backwards. As t runs from 0 to 1, the point f(t) runs through the interval [a, b]. The problem
one runs into, however, is that f need not be smooth. But this obvious attack runs into a
serious diculty. But if we argue as in (5), we run into the integral....., which is meaningless
as it stands. .....where E runs over runs through) the family B.
S
s Unlike Bells method, Halls does not use transnite induction. To show the greater simplicity
of our method over Browns, let us..... Smiths theorem [without the] = the Smith theorem
a theorem of Smiths [= one of Smiths theorems] But we have....., which, by another
theorem of Kimneys, is more than enough to guarantee that P gives A outer measure 1.
Gausss remarkable theorem on..... = the remarkable theorem of Gauss on..... [Adding s is
recommended even after names ending in s, z or x, except ancient ones: Pythagoras theorem;
however, the forms like Jones theorem are also correct.] Feerman and Steins theorem = the
Feerman-Stein theorem [In the plural of abbreviations and numerals, the apostrophe is best
omitted: ve 1s, PDEs, in the 1980s. However, it should sometimes be used to avoid confusion:
There are three is in row k.]
sacrice [see also: cost, price, expense] This extension retains control on..... at the sacrice of
losing some control on..... [Not: loosing]
sake for the sake of simplicity = for simplicity For the sake of clarity, we shall indicate in what
follows to which space X belongs. It seems preferable, for claritys sake, not to present the
construction at the outset in the greatest generality possible.
same [see also: identical] We use the same trick as Boas as Boas does/that Boas does).
.....with the same symbols as are) used in..... The quiver Q
1
is the same as Q but with x
deleted. Then the same argument as in Theorem 5 applies to show that L(R) fails to be
amenable. This is the same as saying that..... Since the code is linear, showing that it is
3-error-correcting is the same as showing that..... But if E is not reexive orwhat is the
samew is weak, then..... But in fact we get the same thing if we consider all maps into S.
Thus A-equivalence and B-equivalence are the same thing. Now G can be handled in much
128 JERZY TRZECIAK
the same way. The same applies to D. So s can be thought of as q with F
q
extended but X
q
left the same.
satisfactory [see also: suitable, appropriate, good, t] Fortunately, there is a very satisfactory
solution to this problem, due to Vermes. It will eventually appear that the results are much
more satisfactory than one might expect. On the whole, the solution can hardly be considered
satisfactory.
satisfy [,= verify; see also: obey] Here a and b are chosen to satisfy (2). [Not: to verify (2)]
This forces f to satisfy (6). The function of Lemma 2 can be made to satisfy (6). The
operator P satises essentially the same inequality as F does. However, A fails to satisfy (3).
Then F need not satisfy (2). [= F does not necessarily satisfy (2).] The products F
i
G
i
are
very close to satisfying (1).
save To save space later on, all modules are given in the form..... Taking J minimal such that
this inequality holds we get a saving of 1 in the codimension.
say [see also: assert, state, tell] When n = 0, (7) just amounts to saying that..... This is
the same as saying that..... To say that A is totally disconnected means that..... Thus f is
bounded, and (1) says that f(a) = 0. This says (roughly speaking) that the real part of g
is..... Let us state once more, in dierent words, what the preceding result says if p = 1. We
now exploit the relation (15) to see what else we can say about G. We cannot hope to say
anything about the structure of each isotropy factor as a system in its own right. If we adjoin
a third congruence to F, say a b, we obtain..... In this case it is advantageous to transfer the
problem to (say) the upper half-plane. Let D be a disc (with centre at a and radius r, say)
in C. Such cycles are said to be homologous (written c c

). [Not: are said homologous]

The usefulness and interest of this correspondence will of course be enhanced if there is a way
of returning from the transforms to the functions, that is to say, if there is an inversion formula.
scale This gure is drawn to a scale of one to ten.
scenario Concentrating on the worst-case scenario, we require D > 0.
scheme [for sth; to do sth; see also: procedure, strategy] Choose S
k
according to the following
scheme.
scope [see also: range, extent] This example falls within the scope of Coxs theorem. It is
beyond the scope of this paper to give a complete treatment of.....
scrutiny [see also: examination, inspection, look] Neighbourhoods of points in these spaces
appear at rst glance to have a nice regular structure, but upon closer scrutiny, one sees that
many neighbourhoods contain collections of arcs hopelessly folded up.
search 1 The search will succeed provided there is.....
2 [see also: look for, seek] Given f, we search for g with properties (1) and (2). By symmetry
considerations, it is sucient to search over a region in which.....
second [see also: other] the second largest element the second last row = the last but one
row the second author = the second-named author To guarantee Q(3), we use MA a second
time to control the perfect sets. A second technique for creating new triangulations out of old
ones is central retriangulation.
MATHEMATICAL ENGLISH USAGE 129
section [,= paragraph] The proof is given in Section 3. [Not: in the Section 3] Choose in
accordance with Section 8. See the simplied account in [2, Section 4]. .....(see the denition
at the beginning of Section 2) In closing this section we take up a result which will play a
pivotal role in the characterization of.....
see [see also: note, observe, view, nd out, discover, grasp, realize, recognize] Since..... we see
that..... We shall be interested in seeing whether..... It is easily seen that..... The conclusion
in (2.3) can be seen by observing that..... This is easily seen to be an equivalence relation. We
saw this in Section 2. We now exploit the relation (15) to see what else we can say about G.
seek [see also: look for, search, try] Here one seeks to minimize P(x) over the class of feedback
controls. In particular, integral curves evolve continuously, and we should seek to represent
them using a measure which reects this continuity in some way. In applications of Theorem 1,
we are usually seeking a lower bound for f(E). We are seeking to verify the conjecture that
every Banach sequence algebra A that is a Ditkin algebra is PAA. Thus f is the function
sought. Therefore, the system (5) has a solution of the sought-for type.
seem [see also: appear] At rst glance Lemma 2 seems to yield four possible outcomes. The
result above seems not to be a consequence of previous results. This approach does not seem
to generalize to arbitrary substructures. In our next theorem, we state a characterization of.....
which does not seem to have been noticed previously. However, it does not seem to have been
observed that..... It seems appropriate to mention in passing the corresponding formula for the
cohomology of B
n
. It seems likely that their results can be extended to..... It seems plausible
that..... but we have been able to establish this only in certain cases. It seems preferable, for
claritys sake, not to present the construction at the outset in the greatest generality possible.
It may seem strange to dene 0. = 0. There seems to be no simple formula for .....
seemingly [see also: apparently] Our rst result generalizes (8) by exploiting some general facts
seemingly overlooked by the aforementioned authors. The approach in [GT] provides a unied
way of treating a wide variety of seemingly disparate examples.
segment Then C lies on no segment both of whose endpoints lie in K.
select [see also: choose, pick] Where there is a choice of several acceptable forms, that form
is selected which..... As we shall see, it is crucial that F is positive for the A-decomposition
selected.
selection [see also: choice] a selection of open problems
self-contained The proof is mainly included to keep the exposition as self-contained as possible.
The proof is self-contained, without reliance on a computer-algebra system.
self-evident There is a self-evident notion of a map between ltered spaces.
seminal [see also: important] Many of these results are known, and indeed they go back to the
seminal paper of Dixmier [D] of 1951.
send [see also: map] Any map either has a xed point, or sends some point to its antipode.
sense [see also: meaning] We remark at the outset that this formula makes sense, because.....
Of course, a literal interpretation of (1) makes no sense. Nevertheless, it might be possible
to make sense of (2) even for non-injective V by considering a multi-valued operator Z. It
is therefore reasonable that the behaviour of p should in some rough sense approximate the
130 JERZY TRZECIAK
behaviour of q. The above denition was rst given in [D]; care is required because this term
has been used in a slightly dierent sense elsewhere. in the Boolean algebra sense We know
that (2) holds in the sense of L
2
convergence. Then F is functorial in the sense that.....
Homogeneous spaces are, in a sense, the nicest examples of Riemannian manifolds.
sensitive [to sth] The results are rather sensitive to the value of the recovery rate .
separate Thus N separates M into two disjoint parts. Such g are linearly dense in C(K) and
hence separate the points of M(K).
separately We estimate the two terms separately. Thus we may count the contribution to the
periodic points from each prime separately. It is also clear that there are extensions to....., but
they do not seem to be worth the eort of formulating them separately. The preceding proof
contains a result which is interesting enough to be stated separately.
sequel In a sequel to the present article, we shall consider..... [= in another paper to be published;
do not write: In the sequel when you mean: In what follows.]
sequence Recall from Theorem 3 that there is a sequence (a
n
) of elements of U that is conal
in M. Altering nitely many terms of the sequence (u
n
) does not aect the validity of (9).
Let (a
n
) be the sequence of zeros of f arranged so that [a
1
[ [a
2
[ ..... Extend this sequence
of numbers backwards, dening N
1
, N
2
and N
3
by..... Then the sequence (8) breaks o in
split exact sequences. Thus the long exact sequence breaks up into short sequences. The exact
sequence ends on the right with H(X). The proof proper [= The actual proof] will consist of
establishing the following statements in sequence. the all-one sequence
series There has since been a series of improvements, of which we briey mention the work of
Levinson.
serious But this obvious attack runs into a serious diculty. Unfortunately, there is a simple
but serious error in the nal step of the proof.
serve This serves to simplify the construction of..... The two examples, E
1
and E
2
, dier by
only a single sequence, e, and they serve to illustrate the delicate nature of Theorem 2. The
same denition serves for f in F. Note that (2) serves as the denition of its left side.
set 1 [see also: collection, family] Let Q denote the set of positive denite forms on V (including
imprimitive ones, if there are any). [Not: the set of the positive denite forms] We take M
to be the set of points in V which map to a point t in A. Here the interesting questions are
not about individual examples, but about the asymptotic behaviour of the set of examples as
one or another of the invariants (such as the genus) goes to innity.
2 [see also: put, establish, dene] Now F is dened by setting F(z) =..... Setting equal to
in Corollary 3, we obtain..... We claim that, by setting w to zero on this interval, the value
of F(w) is reduced. What sets the case n = 5 apart is the fact that homotopic embeddings in
a 5-manifold need not be isotopic. By modifying the technique set out [= presented] in [3], we
obtain..... In Section 2 we set up notation and terminology. [= prepare] On TK we set up
the symplectic structure induced by the metric. [= introduce]
setting [see also: context, set-up] In outline, the argument follows that of the single-valued
setting, but there are several signicant issues that must be addressed in the n-valued case.
The proof makes use of many of the ideas of the general case, but in a simpler setting.
MATHEMATICAL ENGLISH USAGE 131
[3] contains an extension of Proposition 2 to the setting of nitely additive set functions. The
point is that the operator is now much easier to analyse than is the case in the original setting
of the space B. It seems that the relations between these concepts emerge most clearly when
the setting is quite abstract, and this (rather than a desire for mere generality) motivates our
approach to the subject. Here the functional analytic tools required are simpler, but this easier
setting allows us to develop some methods without undue worry about technicalities.
settle [see also: decide, determine, conclude] This argument also settles the case of K = .
In this paper, we settle that question for normal spaces. The only references known to the
authors are [A] and [V], where the case A = L(E) is settled in the negative.
set-up [see also: context, setting] In the set-up of condition (H), let.....
several [see also: some, few, many, couple, number, dierent, variety] We take advantage of
this fact on several occasions, by not actually specifying the topology under consideration.
A model for analysing rank data obtained from several observers is proposed. The present
proof is so arranged that it applies without change to holomorphic functions of several variables.
[Not: of many variables] The argument is a variant of one in [5] and has been used several
times since. Where there is a choice of several acceptable forms, that form is selected which.....
This argument is invalid for several reasons. There are several cases to consider:.....
shape [see also: form] This shape bears a striking similarity to that of..... The set A is roughly
triangular in shape.
share [see also: enjoy, have] Certain other classes share this property. It is a pleasure to thank
R. Greenberg for bringing his criterion for..... to our attention, and for generously sharing his
ideas about it.
sharp [,= strict] The estimate is sharp, as the following example shows. A natural question is
how sharp the bounds given in Theorem 6 are.
sharpen [see also: improve, rene, strengthen] The following lemma is a sharpening of well-
known results concerning indecomposable ultralters.
shed Our result sheds a new light on the problem of.....
short [see also: brief, brevity, abbreviation, shorthand, fail] The method falls short of providing
an explicit formula for the index. We write G = FHF
1
for short. [= for brevity; not: We
write shortly] In short, F preserves the algebraic structure of M. [= To sum up briey]
Our main results state in short that MEP characterizes type 2 spaces among reexive Banach
spaces.
shortcoming [see also: disadvantage, weakness, fail] A shortcoming of our method is the in-
ability to compare three or more progressions.
shorthand [see also: abbreviation, brevity, short] Here k is shorthand for k([x[). Let us adopt
the shorthand F; = FM
i
N
i
.
shortly [,= briey, ,= for short; see also: moment] .....where F will be dened shortly. [= in a
moment] We shall prove this theorem shortly, but rst we need a key lemma. The benet of
formulating our notion of isomorphism section as above will become clear shortly. However,
shortly after learning about Wieners work, P. Levy found a more elementary argument.
132 JERZY TRZECIAK
should It is therefore reasonable that the behaviour of p should in some rough sense approximate
the behaviour of q. It should be noted that..... It should be possible to prove this directly
by studying the xed point set of the action of G. It should now be clear that A = B. One
unusual feature of the solution should be pointed out. The diculty is that it is by no means
clear what one should mean by a normal family. Denote the largest of those w
i
by w
p
. Should
there be no such w
i
, let w
p
= 0. [= If there is no]
show [see also: prove, establish, demonstrate, evidence, reveal, reect, represent, display, exhibit,
indicate] The series can easily be shown to converge. To show that..... is equivalent to showing
that..... Since the code is linear, showing that it is 3-error-correcting is the same as showing
that..... Examples are given to show that..... As shown by Faraut, this expansion is convergent.
This is true for any family connecting the two ows shown. This space of curves also shows
up in the theorem of Meyer.
shrink By shrinking A to a point, we obtain.....
side [see also: member] There is no condition relating the sections on one side of N to those on
the other side. Integrating both sides over X shows that..... the right side = the right-hand
side This question was answered negatively in [5]. However, on the positive side, Davies [5]
proved that..... Our focus now will be on one-sided averages. [Do not write: On the other
side if you mean: On the other hand.]
sight [see also: glance] at rst sight
sign The terms of the series (1) decrease in absolute value and their signs alternate. equal up
to sign the minus sign
signicance [see also: importance, role, relevance, interest] The signicance of..... is that.....
The signicance of this fact for our purposes is captured by Corollary 3. In the study of
innite series

a
n
it is of signicance whether the a
n
approach zero rapidly. This has deeper
signicance than one might rst realize.
signicant [see also: considerable, important, substantial] This can result in a signicant loss
of smoothness. It is important to recognize that there is a signicant subclass of vector elds
for which the ambiguity alluded to above disappears. In all our analysis, only the order of
magnitude of P will be signicant. Their result gives no information when k is large, whereas
(5) is signicant regardless of the size of k.
signicantly [see also: considerably, greatly, substantially] The other values are signicantly
smaller than x. The pressure increases are signicantly below those in Table 2.
signify [see also: mean] Under those conditions, what does the sum on the left hand side of (8)
signify?
similar [to sth; see also: like, reminiscent, resemble] It is in all respects similar to matrix
multiplication. [Not: similar as] In the course of writing this paper we learned that P. Fox
has simultaneously obtained results similar to ours in certain respects. The proof is similar in
spirit to that of [8]. Analysis similar to that in Section 2 shows that..... Similar arguments to
those above show that..... This says that I is no longer than the supremum of the boundary
values of G, a statement similar to (1). A similar result holds for..... in a similar fashion in
an exactly similar way
MATHEMATICAL ENGLISH USAGE 133
similarity [see also: resemblance] This shape bears a striking similarity to that of..... Note
the similarity with Lusins theorem.
similarly [see also: likewise] Then F is similarly obtained from G. Similarly to [4], we rst
consider the nondegenerate case. [Or: Just Much) as in [4]; not: Similarly as in [4]] Here
we consider a dual variational formulation which can be derived similarly to that for the sandpile
model. Each A(n) corresponds to an element A

(n) in V , and similarly for B(n).

simple [see also: easy, straightforward, direct] Unfortunately, there is a simple but serious error
in the nal step of the proof. Here is a simple direct proof. This strikingly simple proof
was discovered by J. Dixon. It is useful to consider some rather simple examples to gain some
intuition. Although the idea is simple, its implementation is complicated by the fact that.....
In fact, the computation for A becomes somewhat simpler. We remark that this proof seems
conceptually and calculationally simpler. We rst prove the (rather simpler) Theorem 7, by
eecting a quite general reduction of the problem to the study of certain isotropy factors. The
simplest example of this is furnished by.....
simplicity [see also: ease] For simplicity of notation, we use the same letter f for..... For
simplicity, we suppress the explicit dependence on x in the notation. One can also study these
equations on manifolds, but we stick to R
n
for simplicity.
simplify To simplify the writing, we take a = 0 and omit the subscripts a. This simplies to
f = g for x = 1. For this choice of , and with u = z = s, the expression (5.3) simplies
greatly. It simplies the argument, and causes no loss of generality, to assume..... This
assumption is useful for simplifying proofs. Neither is the problem simplied by assuming
f = g. See the simplied account in [2, Section 4].
simply [see also: just] If we simply mimic the standard proof of..... we are led to..... If M
is empty we write simply f g. The rank of appearance (or simply the rank) of an integer
in the sequence U(P) is the least positive integer n such that..... The lower limit is dened
analogously: simply interchange sup and inf in (1). Then G is simply g with its periodic string
read backwards. The reason for preferring (1) to (2) is simply that (1) is manifestly invariant.
[Note the double r in preferring.] The presence here of the direct summand H is simply to
prevent A from having disconnected spectrum.
simultaneously [see also: time] These two possibilities cannot arise simultaneously. Clearly,
A is thereby put in one-to-one correspondence with B. In the course of writing this paper we
learned that P. Fox has simultaneously obtained results similar to ours in certain respects.
since 1 [see also: as, because] Since f is compact, we have Lf = 0. Since f is compact, it
follows that Lf = 0. [Not: Since f is compact, then Lf = 0.] [In sentences beginning with
Since, it is sometimes unclear which part is the premise, and which is the conclusion: instead of
writing: Since A = B, B = C, C = D, A = C, write e.g.: Since A = B and B = C, and since
moreover C = D, it follows that A = C.]
2 The argument is a variant of one in [5] and has been used several times since. There has
since been a series of improvements, of which we briey mention the work of Levinson.
single [see also: individual, particular, specic, unique, only] 1 This will be proved by showing
that H has but a single orbit on M. Each row of A has a single 1 and the rest of the entries 0.
The two examples, E
1
and E
2
, dier by only a single sequence, e, and they serve to illustrate
134 JERZY TRZECIAK
the delicate nature of Theorem 2. Can E consist of a single point?
2 With this denition of a tree, no vertex is singled out as the root.
situation [see also: circumstances, state] We do not know whether or not Q(R) = R in this
situation. Rather than discuss this in full generality, let us look at a particular situation of this
kind. Suppose, to look at a more specic situation, that..... The situation is quite dierent
if we replace H(U) by certain subclasses. If A = B, how does the situation dier from the
preceding one? As opposed to the situation considered in [5], the functions used here are.....
There are many situations where this occurs naturally. We shall encounter similar situations
again, and shall apply convergence theorems to them without further comment. Here are some
other situations in which we can draw conclusions only almost everywhere.
size [see also: magnitude] Clearly, F has size p. All inputs of size n are equally likely to
occur. Their result gives no information when k is large, whereas (5) is signicant regardless
of the size of k. The aim of this article is to study the relationship between the size of A, as
measured by its diameter, and the extent to which A fails to be convex. Then P is the product
of several integer factors of about x
n
in size.
sketch [see also: outline] 1 A brief sketch of the reasoning is given below.
2 We sketch the proof of the easy half of the theorem.
skip We do this in the rst section, which the reader may skip on a rst reading.
slant area shaded with slanting lines
slight [see also: little, small, minor] However, the chance of success is very slight. However,
a slight strengthening of the hypotheses does give us a regular measure. We write = .....,
which is a slight modication to the previous version of .
slightly [see also: little, somewhat] We have to change the proof of Lemma 3 only slightly.
Unfortunately, the notation from number theory slightly conicts with the notation from
probability theory. Slightly rening a result of [8], Davenport proved..... The estimates (5)
then follow by slightly increasing the value of m. They are all slightly dierent. Note that
(2) is a slightly weakened version of the Polya inequality. The motivation for the results of this
section is the following result of John (paraphrased slightly to suit our purposes).
small [see also: little, minor, slight, below] ....., where C can be made arbitrarily small by
taking..... [Not: arbitrary small] for all suciently small h However small a neighbourhood
of x we take, the image will be..... [= No matter how small] The other values are signicantly
smaller than x. Therefore, F is not no) smaller than G. Thus A is the smallest algebra
with this property. There is a smallest algebra with this property. Then B contains a unique
element of smallest norm.
so [see also: consequently, hence, thus, therefore, way] Indeed, it is routine to verify that the
index so constructed is independent of the choices made. The constants are so adjusted in (6)
that (8) holds. .....where C is so chosen chosen so) that..... The control problem is to choose
an investment strategy so as to minimize..... The rst of these was suggested by J. Serrin, who
showed how to modify my earlier treatment of..... so as to obtain stronger results with no extra
eort. The ordered pair (a, b) can be chosen in 16 ways so as not to be a multiple of (c, d).
Thus A is compact, and hence so is B. Consequently, A is compact. Therefore, so is B.
Since R is polynomial in x, so also is P. We have been working under the assumption that.....
MATHEMATICAL ENGLISH USAGE 135
Suppose now that this is no longer so. If this is so, we may add..... If this is not so, a linear
fractional transformation will make it so. Then F is bounded but it is not necessarily so after
division by G. We will want to prove that..... To do so, we will need to dene..... A geodesic
which meets bM does so either transversally or..... This shows that the sequence (1) is bounded
below, and so the denition of L(f) is meaningful. Then a = b, so that f is not injective.
solely [see also: alone, only, single] Since the entire argument is based solely upon assumption
(6.1), the conclusion of the theorem must hold.
solution [see also: answer, explanation] This equation has a unique solution for every p. To
guarantee existence and uniqueness of a solution to (3), the function g must be..... Fortunately,
there is a very satisfactory solution to this problem, due to Vermes. This observation prompted
the author to look for a more constructive solution. It seems that the solution of Problem 1 is
still out of reach.
solve [see also: resolve] We can solve the resulting equations successively for c
1
, ..., c
n
. We
remark that Bennett and the third author completely solved equation (4) in the case when.....
If y is a solution, then ay also solves (3) for all a in B. This easily allows the cases c = 1, 2, 4
to be solved. This problem is still far from being solved. The MR algorithm takes n steps to
solve the problem.
some [see also: few, several, couple, number] Each f lies in zA for some A. Note that some
of the X
n
may be repeated. Some of the isomorphism classes above will have a rank of 2.
Some such diculty is to be expected. The theorem implies that some nite subcollection of
the f
i
can be removed without altering the span. It is therefore reasonable that the behaviour
of p should in some rough sense approximate the behaviour of q.
something Strictly speaking, we should write something like a(l, m, n) to reect the dependence;
we shall rely upon context instead. Specically, one might hope that a clever application of
something like Choquets theorem would yield the desired conclusion.
sometimes [see also: occasionally, occasion] This is sometimes expressed by saying that.....
The name Riesz theorem is sometimes given to the theorem which asserts that..... The
sequence a
n
is what is sometimes called a recovery sequence for v.
somewhat [see also: little, slightly, fairly, rather] Our results in this section can be rened
somewhat by considering..... In fact, the computation for A becomes somewhat simpler. This
is somewhat surprising since..... We shall discuss this again at somewhat greater length in
Section 2.1. Fox obtained a somewhat better result:.....
soon [see also: once] This will be proved as soon as we show that.....
sophisticated [see also: complex, intricate, rene] To recover Wieners famous result that
Brownian paths are continuous, one needs to use more sophisticated reasoning.
sort [see also: kind, type] This result may be thought of as a sort of regularity theorem. Our
aim here is to give some sort of functorial description of K in terms of G. This sort of proof
will recur frequently in what follows. This sort of tacit convention is used throughout Gelfand
theory.
source [see also: origin] An excellent source on these matters is [G75]. Since most of the
results presented are quite classical (the novelty lies in the arrangement, and some of the proofs
136 JERZY TRZECIAK
are new), I have not attempted to document the source of every item. the main source of
inspiration
space 1 There is not space to enumerate them all here. To save space later on, all modules are
given in the form..... a function continuous in space variables Let R
n
be Euclidean n-space.
[Or: the Euclidean n-space]
2 The m points x
1
, . . . , x
m
are regularly spaced t units apart.
speak [see also: talk] Roughly speaking, we shall produce a synthesis of index theory with
Fourier analysis. Strictly speaking, we should write something like a(l, m, n) to reect the
dependence; we shall rely upon context instead. Hence, although the topology of reducts of A
is uniformly controlled, so to speak, by that of A, the model theory of the reducts can be much
wilder. The two polygons P and P

are the two ends of the orbit, so to speak.

special [see also: individual, particular, specic] An important special case is when L is empty.
This abstract theory is not in any way more dicult than the special case of the real line.
specialize For f = g, this specializes to the result of [7]. Now (3), specialized to our case,
becomes..... We now specialize to the situation of Lemma 8. More specialized notions from
Banach space theory will be introduced as needed.
specic [see also: concrete, individual, particular, special, deniteness] Suppose, to look at a
more specic situation, that..... This implies that the local martingale must take a very specic
form. Section 2 contains some specic preparatory material, notably a brief discussion of the
category of virtual vector bundles. To be more specic, consider.....
specically It is smoothness with which we are specically concerned. Specically, choose
A = B = id/dx. Specically, one might hope that a clever application of something like
Choquets theorem would yield the desired conclusion.
specify ....., where each function g is as specied above. Choose a sequence of compact sets K
n
with the properties specied in Lemma 3. The parent and child relations often used with trees
can be derived once a root is specied. In the latter case we may simply adjust F to equal 1
on the Borel set where it falls outside the specied interval. We take advantage of this fact
on several occasions, by not actually specifying the topology under consideration. We will not
normally specify the index set I.
spirit The proof is similar in spirit to that of..... Although the denition may seem articial, it
is actually very much in the spirit of Darbos old argument in [5].
split [see also: decompose, subdivide] Moreover, M splits as K L. Now (1) splits into the
pair of equations..... The proof conveniently splits into two cases. We shall split up K as
follows.
square By repeated squaring to eliminate the radicals in this equality, we obtain.....
stage [see also: degree, extent, step] We are now ready to proceed to the nal stage of our
construction. At this stage we appeal to Theorem 2 to deduce that..... We may require that
the point P lie in one of the trees constructed before or during the ith stage of the induction.
at stage n
stand [see also: state] Here dx stands for Lebesgue measure. [Or: the Lebesgue measure] The
model as it stands does not satisfy the conditions of..... But if we argue as in (5), we run into
MATHEMATICAL ENGLISH USAGE 137
the integral....., which is meaningless as it stands. We make two standing assumptions on the
maps under consideration.
standard [see also: common, customary, familiar, usual, routine] Using the standard inner
product we can identify H with H

. The standard proofs proceed via the Cauchy formula.

A standard verication shows that..... It is a standard fact that..... Let us sketch the proof of
the rst estimate, which is a variation of standard arguments. Although standard, the notion
of a virtual vector bundle is not particularly well known.
standpoint [see also: viewpoint] From the standpoint of linear programming, the above discus-
sion is incomplete in that it throws no light upon the question whether the function F attains
its inmum.
start 1 [see also: beginning, outset] We may thus assume from the start that..... We now full
the promise made at the start of the proof by handling the case of G = S
n
.
2 [with sth; by doing sth; see also: begin, commence, proceed] We start with the observation
that..... Lemma 3 suggests that we start by considering A
1
. In [2], this theorem is made the
starting point of Gelfand theory. curves starting at the origin
state 1 [see also: circumstances, situation, condition] The number of distinct states a processor
may be in is at most n.
2 [see also: assert, say, give, formulate, stand] We end this section by stating without proof
an analogue of..... In order to state these conditions succinctly, we introduce the following
terminology. In our next theorem, we state a characterization of..... which does not seem to
have been noticed previously. The preceding proof contains a result which is interesting enough
to be stated separately. Stated informally, continuous functions of measurable functions are
measurable. We have not required f to be....., and we shall not do so except when explicitly
stated. Hence A has the stated continuity properties. Part (b) follows from (a) on noting
that A = B under the conditions stated. Actually, [3, Theorem 2] does not apply exactly as
stated, but its proof does.
statement [see also: assertion, conclusion, formulation] The following three statements are
equivalent:..... Now, (3) is merely an abbreviation for the statement that..... A statement
equivalent to (c) is that to each w there corresponds a sequence..... Here is a more explicit
statement of what the theorem asserts. The convergence of the sum on the left is of course a
weaker statement than the convergence of (2). It is important that the orders of F and G are
comparable, a statement made more precise by the following lemma. However, (ii) is nothing
but the statement that..... [= only the statement that]
stay [see also: remain, continue] The point is that (5) can stay small because..... If we stay a
xed distance o the critical line, we do not expect Benford behaviour. The Markov chain z
k
takes no account of how long the process stays in V . This path stays in B for all time.
step [see also: stage] The remainder of our work breaks into ve steps. As a rst step we shall
bound A below. For the base step of the induction, consider a vertex t in A. A key step in
obtaining (6) is Jensens inequality. This was one of the major steps in Wieners original proof
of his Tauberian theorem. ....., from which it is an easy step, via Lemma 1, to the conclusion
that..... The next step is to establish..... Going back to the existential step of the proof,
suppose that..... All the computation will be done by one processor in one step. By reversing
the steps above, we see that..... Using (2) and following steps analogous to those above, we
138 JERZY TRZECIAK
obtain..... To be sure we can carry out step n, we need to know that..... The MR algorithm
takes n steps to solve the problem. at each step of the construction
stick [see also: keep, continue] One can also study these equations on manifolds, but we stick to
R
n
for simplicity. All our results can be extended in this way, but we shall stick to considering
P rather than P

.
still [see also: continue, yet] The operators A
n
have still better smoothness properties. The new
X is no more continuous, although it still has norm 1. An ingenious alternative proof, shorter
but still complicated, can be found in [MR]. This is of course still an implicit characterization
of V , since..... It seems that the solution of Problem 1 is still out of reach. Then we can nd
a subsequence (still denoted by a
n
) such that a
n
< 1 for all n.
stipulation [see also: requirement, demand] These extra stipulations are unimportant, but are
given for deniteness.
straight [see also: directly] Then the one and only integral curve of L starting from x is the
straight line l. The boundedness follows straight from the denition of G.
straightforward [see also: direct, easy, simple] Alternatively, it is straightforward to show
directly that..... The proof of (2) is a matter of straightforward computation, and depends on
the relation ab cd = 1. in a fairly straightforward way
strange [see also: unusual] It may seem strange to dene 0. = 0.
strategy [see also: procedure, scheme, method, way] Our basic strategy for proving (1) is
dierent. We start with a brief overview of our strategy. Furthermore, adopting this strategy
considerably eases constructing a coding tree from a linear order. The condition..... can be
improved by employing a strategy similar to that underlying the proof of Theorem 2. The
strategy is much the same as for the proof of Theorem 2. a basic broad/general/overall/viable)
strategy
strength [see also: force] We do not need the full strength of the bounds in (4).
strengthen [see also: improve, sharpen, rene, strong] Theorem 3 can be strengthened by re-
quiring that the ultralter U be additionally conservative. [Note the subjunctive be.] However,
a slight strengthening of the hypotheses does give us a regular measure.
stress 1 [see also: emphasis] In his lectures, he laid great stress on the use of.....
2 [see also: emphasize, underline] Let us stress that c is a term and not a subset of C. We
should stress that this is only one of several versions of the measurable selector theorem, due
variously to von Neumann, Jankow, Lusin and others. It should be stressed, however, that.....
strict Strict inequality [= with a < or > sign] can occur hold) in (8) only if..... The condition
(8) holds with strict inequality.
strictly [see also: properly] This class is strictly larger than..... Then F is strictly less than G.
strike This shape bears a striking similarity to that of..... All three cases bear a striking
resemblance:..... This is in striking contrast to..... The two gures appear strikingly dierent.
stringent [see also: restrictive] Condition (i) becomes more stringent as n increases.
strive Striving for a contradiction, assume that.....
MATHEMATICAL ENGLISH USAGE 139
stroke [see also: go] The idea of the following proof, which yields both (a) and (b) at one stroke,
is due to von Neumann.
strong [see also: strengthen] A stronger topology makes it easier for a given function to be
continuous. In fact, this proof shows that a stronger result holds, namely,..... The rst of
these was suggested by J. Serrin, who showed how to modify my earlier treatment of..... so as
to obtain stronger results with no extra eort.
structure [see also: form] Then G has the structure of a group such that..... Neighbourhoods
of points in these spaces appear at rst glance to have a nice regular structure, but upon closer
scrutiny, one sees that many neighbourhoods contain collections of arcs hopelessly folded up.
This requires us to give M an R-module structure. Note that E can be given a complex
structure by setting..... Thus, the tensor algebra acquires a graded algebra structure. They
dened the concept solely in terms of the norm of the Banach space, deliberately avoiding
any extra structure. The structure of a Banach algebra is frequently reected in the growth
properties of its analytic semigroups.
study 1 [see also: consideration, discussion, examination, investigation, research, exploration,
analysis] Our study grew out of some valuable conversations with Kirk Douglas. Their study
resulted in proving the conjecture for..... Much of the rest of the paper is devoted to a general
study of..... In Section 2, we lay the foundations for a systematic study of..... In the study of in-
nite series

a
n
it is of signicance whether the a
n
approach zero rapidly. In Section 3 we pro-
ceed with the study of..... [= start or continue the study] In [7] the authors undertook a detailed
study of..... a careful comprehensive/close/systematic/comparative/denitive/profound/
thorough/meticulous) study
2 [see also: consider, examine, investigate, explore] One can also study these equations on
manifolds, but we stick to R
n
for simplicity. If one studies the proof of..... it is apparent that
(2) is never used. The aim of this article is to study the relationship between the size of A,
as measured by its diameter, and the extent to which A fails to be convex. This is part of
a larger project to study the Galois groups of periodic points of arbitrary polynomial maps.
This functor is much studied in [Ab] and [Ce]. These three results lead to several illuminating
pieces of information about the (insuciently studied) Berger property in general spaces.
subdivide [see also: split] Each ray will subdivide into a pair of dynamically distinct pieces.
subject 1 [see also: theme, topic, matter, point, issue] This subject is treated at length in
Section 2. See also [3], where functions of exponential type are the main subject. We cannot
survey this whole subject here. It seems that the relations between these concepts emerge most
clearly when the setting is quite abstract, and this (rather than a desire for mere generality)
motivates our approach to the subject.
2 Choose f in G subject only to the condition that Lf = 0. They established the Hasse principle
subject to a rank condition on the coecients. Take N to be a family of normal measures in
P(X) such that N is maximal subject to the condition that the supports of the measures in the
family are pairwise disjoint. The location of the zeros of a holomorphic function in a region
is subject to no restriction except the obvious one concerning the absence of limit points in .
subordinate a partition of unity subordinate to the covering U
i

subscript For brevity, we drop the subscript t on h

t
. We drop the subscript when confusion
is unlikely. To simplify the writing, we take a = 0 and omit the subscripts a. .....where the
140 JERZY TRZECIAK
subscript r denotes radial dierentiation.
subsequence Then we can nd a subsequence (still denoted by a
n
) such that a
n
< 1 for all n.
Then we can nd a subsequence (not renamed) such that a
n
< 1 for all n. Thus every
subsequence contains a further subsequence converging weakly to some limit. Passing to a
subsequence if necessary, we can assume that.....
subsequent [see also: follow] Theorem 2 will form the basis for our subsequent results.
subsequently [see also: next, then, below] Here and subsequently, M denotes..... We note in
passing that Fox has subsequently improved Barness result by showing that.....
substance [see also: essence, point, nature] What the theorem is saying in substance is that.....
We should expect that..... This expectation is given substance by the following result.
substantial [see also: considerable, signicant, extensive, wide] This proof will require sub-
stantial changes to cover the case of..... A substantial no-mans-land persists in which neither
alternative has been proved. Our proofs make substantial use of classical topology of the plane.
substantially [see also: considerably, signicantly, largely] The number of..... has increased
substantially [= considerably] in recent years. The purpose of this paper is to expand substan-
tially the class of maps for which the index can be computed. The level of..... has remained
substantially [= largely, essentially] unchanged for many years.
substantiate This claim, however, might be dicult to substantiate. [= prove]
substitute [see also: put, replace] We can substitute a subspace A of X for the set 0, 1.
Substitute this value of z into in) (7) to obtain.....
subsume Most of these phenomena can be subsumed under two broad categories. [= included
in]
subterfuge [= trick] The analogous construction when A is not ane entails [= requires] the
following subterfuge.
subtle [see also: delicate, ne, minor] Although these proofs run along similar lines, there are
subtle adjustments necessary to t the argument to each new situation.
subtlety There is a slight subtlety, since A is not exactly equivalent to B.
subtract [not: substract] Since Z is a nite set, we may continue subtracting suitable scalar
multiples of the x
i
from x. By subtracting a constant if need be, we may assume.....
succeed [in doing sth; see also: work] The search will succeed provided there is..... We suc-
ceeded in proving (4) for amenable groups. [Not: succeeded to prove]
success They have a 90 per cent chance of success. However, the chance of success is very
slight.
successful Mary Lane deserves our special thanks for her part in bringing this volume to a
successful completion.
successfully The same kind of approach has been taken successfully to determine all regular
maps of various small genera.
MATHEMATICAL ENGLISH USAGE 141
successive [= consecutive] Successive vertices on a path have alternating labels. an interval
of successive integers
successively [= consecutively] We can then successively determine F in the other components.
We can solve the resulting equations successively for c
1
, ..., c
n
.
succinct [= concise; see also: brief] a succinct account of the theory
succinctly [= concisely] To state these conditions succinctly, we introduce the following termi-
nology.
such Assume that such a g exists. Theorem 1 can be used to bound the number of such T.
There is a map such that..... [Not: There is such a map that] One such mapping is the
function f given in..... There are few, if any, other signicant classes of processes for which such
precise information is available. Some such diculty is to be expected. For general linear
operators, there is not such an extensive functional calculus as there is for self-adjoint operators.
a shorter such proof every such map many such maps Here the interesting questions are
not about individual examples, but about the asymptotic behaviour of the set of examples as
one or another of the invariants (such as the genus) goes to innity. This convention simplies
the appearance of results such as the inversion formula. We observe a prolonged rise in prices
such as occurred in the late 1960s. Our presentation is therefore organized in such a way that
the analogies between the concepts of topological space and continuous function, on the one
hand, and of measurable space and measurable function, on the other, are strongly emphasized.
Such was the case in (8). We can continue to pick elements of B as above. But there are
only nitely many such, a contradiction.
suce [see also: sucient, enough] From Proposition 2, and in view of Theorem 3, it suces
to show that..... To tell whether a labelled partial order is a skeleton, it suces to look at its
substructures of size at most three. For our purposes, suce it to say that if A is close to B,
then.....
suciency We now prove suciency. [Or: the suciency]
sucient [see also: enough] Neither (1) nor (2) alone is sucient for (3) to hold. This is clearly
sucient to make the computation for T. However, (5) is sucient to guarantee invertibility
in A. By symmetry considerations, it is sucient to search over a region in which.....
suciently [see also: enough] for all suciently small h for n suciently large
suggest [doing sth; see also: indicate, propose, recommend] This suggests that A can exist.
Unfortunately, the details of the calculations were omitted, and there is some doubt on whether
the result is correct since our analysis suggests that P
2
must vanish to third order; the presence of
L
2
is also suspect. Lemma 3 suggests that we start by considering A
1
. Some other problems
suggest themselves in the topological setting. This suggests the introduction of the dierential
operator A =..... This suggests a question: under what conditions is it true that.....? The
above construction suggests investigating the solutions of..... The rst of these was suggested
by J. Serrin, who showed how to modify my earlier treatment of..... so as to obtain stronger
results with no extra eort. [Do not write: He suggested me to study this if you mean: He
suggested studying this or He suggested I should study this.]
suggestion The author thanks the referee for his helpful suggestions concerning the presentation
142 JERZY TRZECIAK
of this paper. We are grateful to the referee for a number of helpful suggestions for improvement
in the article. At the suggestion of the referee, we consider some simple cases.
suit [see also: t, convenient] This denition best suits our purpose of showing that..... The
motivation for the results of this section is the following result of John (paraphrased slightly to
suit our purposes). A detailed exposition, more suited to the purposes of the present article, is
given in [9]. It became clear that the Riemann integral should be replaced by some other type
of integral, better suited for dealing with limit processes.
suitable [see also: appropriate, convenient, good, correspond, relevant, t, adapt] By suitable
translation of variables in (5), we may arrange that k =..... [Not: By adequate translation]
It suces to take a suitable translate of U. Having established (1), one might be tempted
to try to extend this result to general p through the choice of a suitable ideal B. Since Z is a
nite set, we may continue subtracting suitable scalar multiples of the x
i
from x.
suitably [see also: appropriately] Here Y is a Poisson variable suitably chosen to approximate X
in distribution. We nish by mentioning that, suitably modied, the results of Section 2 apply
to the AP case. with N being a suitably large constant in a suitably small neighbourhood
of zero
sum 1 Hence F is the sum of an injective module and a projective module. Here p
2
(r) is the
sum of the squares of the divisors of r. Every F is a sum of irreducible elements. Here F is
the sum of a collection of..... The sum is taken over all a dividing p.
2 [see also: add] Summing (2) and (3), we obtain..... Keep only those vertices whose coordi-
nates sum to 4. [= add up to 4]
summarize [,= resume] We summarize some of its main properties, borrowing from the elegant
discussion in Hensons article. Theorems 2 and 3 may be summarized by saying that.....
Summarizing, whenever n 4, we have shown that necessarily p < 5.
summary In summary, if F is..... In particular, there is a summary of what is known about
polarized pairs with small genus.
superuous [see also: redundant] We now prove that the positivity assumption is in fact
superuous.
superior However, (5) is superior to (4) only when n is cubefree.
supply [see also: give, furnish, oer, provide, yield] This equation supplies the key to the
proof of Theorem 2. It is obvious that the above theorem supplies an algorithm to eectively
recognize whether SP is in A. The statement does appear in [3] but there is a simple gap in
the sketch of proof supplied.
support 1 a function with compact support. We now prove that f cannot have compact support
unless f = 0 a.e. This research was initiated when the rst author was visiting the University
of Alberta in the summer of 2008; the nancial support and kind hospitality are gratefully
acknowledged.
2 Research partially supported by NSF Grant No. 23456. Then F is supported in [z[ 1.
Some partial evidence to support this conjecture is discussed in [3]. We oer numerical
evidence to support a conjecture that there exist innitely many primes of this type.
MATHEMATICAL ENGLISH USAGE 143
suppose [see also: assume, presume] Suppose, to look at a more specic situation, that.....
Suppose that, contrary to our claim,..... We now show that A is closed. Suppose that, on the
contrary, there is an x..... Suppose towards a contradiction that..... We now prove..... Indeed,
suppose otherwise. Then..... Suppose for the moment that q = 1, so that = 1.
suppress For simplicity, we suppress the explicit dependence on x in the notation. If G is clear
from context, then we suppress reference to it in the notation. We will work exclusively in
the category of standard Borel probability spaces, and so will often suppress mention of their
sigma-algebras.
supremum [see also: maximum] Taking supremum over all g such that..... we get..... The
measure of A is the supremum of the measures of its compact subsets.
sure To be sure we can carry out step n, we need to know that.....
surely [see also: certainly] The following easy lemma is surely folklore. We now prove a simple
fact about semigroups: this is surely a folklore result. These upper bounds are too large to be
useful in computer calculations in general, but the ideas in the proofs will surely contribute to
better bounds in the future. A formula like (3) surely deserves some explanation.
surpass [see also: exceed, outnumber] Prewar levels of production were surpassed in 1929.
surprise 1 Much to our surprise, even for k = 1 the problem proved quite dicult. It should
come as no surprise that a condition like a
i
,= b
i
turns up in this theorem.
2 We were surprised to nd out that..... at nding out that.....) This is somewhat surprising
since.....
surprisingly Not surprisingly, zero sets of sections of such an L are called ample divisors. The
surprisingly neat answer is that the functions (2) span C(I) if and only if G(a) = 0.
survey [see also: overview] 1 For an extensive survey of....., see..... For a survey of what is
known to date, see [G]. A survey of the research on f
n
(x, y) up to 2007 (most of it dealing
with the case n = 1) was given in [3]. The survey article [5] by Diestel contains a wealth of
information about the DunfordPettis property.
2 We cannot survey this whole subject here. Other situations in dynamics where the p-adic
numbers come up are surveyed in [W].
suspect 1 Indeed, there is reason to suspect that diculties could increase with increasing n.
We suspect that our methods should extend to oer at least some description of characteristic
factor-tuples for larger numbers of commuting transformations. While we suspect that the
methods of the present section can be brought to bear on this question also, we will not pursue
this analysis in detail here.
2 Unfortunately, the details of the calculations were omitted, and there is some doubt on whether
the result is correct since our analysis suggests that P
2
must vanish to third order; the presence
of L
2
is also suspect.
symbol [see also: denote, write, notation] Here A, B and C all belong to the same class, which
we represent by the symbol P
2
. We shall use the symbol to represent intersection. We call
F and G equivalent (in symbols F G) if.....
symmetry The symmetry of G about 0 gives..... By symmetry considerations, it is sucient
to search over a region in which..... a crystal with hexagonal symmetry
144 JERZY TRZECIAK
synthesis Roughly speaking, we shall produce a synthesis of index theory with Fourier analysis.
system The extended real number system is R with two symbols, and , adjoined.
Therefore, the system (5) has a solution of the sought-for type.
systematic In Section 2, we lay the foundations for a systematic study of.....
systematically There are quite a number of cases, but they can be described reasonably sys-
tematically.
T
table Direct verication establishes the other four cases (see the table below).
tabulate We tabulate the outcome for n 10; in particular, the column headed R lists R(n)
truncated to three decimal places. The smallest (most delicate) values are tabulated below.
tacit This sort of tacit convention is used throughout Gelfand theory.
tacitly We tacitly assume that..... Of course, it is tacitly understood that it is this measure
that is really under discussion.
take [see also: address, consider, deal, adopt] One can, for example, take A to be the rationals
in X. The parameter interval was here taken to be (0, 1). Let us take B as our range space.
Take as base for a topology on X the sets of the form..... The constant C may be taken as double
the constant appearing in (5). Take for H the set..... Taking y = x we get..... .....where the
sup is taken over all intervals I. It turns out that this is independent of the representations
taken (as long as they are faithful). The map f takes a to f(a). The map f takes the value
1 for t = 1. The function g takes its maximum at x = 5. The MR algorithm takes n steps
to solve the problem. Following the same lines we nd that it takes k prolongations to get an
immersed curve. We take the same approach as in [3]. This procedure can be extended to
take care of any number of terms. Let us now take a look at the class N, with the purpose of
determining how..... A linear transformation takes us back to the case in which..... Then (5)
takes on the form..... [Or: takes the form] Here f takes over the role of the time parameter.
The rst 15 chapters should be taken up in the order in which they are presented, except for
Chapter 9, which may be postponed. First we take up the trivial case h = 0. In closing this
section we take up a result which will play a pivotal role in the characterization of.....
talk 1 [= lecture] He gave an interesting talk on.....
2 [see also: speak] When we talk of a complex measure, it is understood that (E) is a complex
number.
task [see also: objective, aim] We can accomplish both tasks by showing that..... The task is
now to obtain..... Solving it in a general setting is a complicated task. Although individually
these systems can still be quite complicated, a possibly more tractable task is to describe their
possible joint distributions.
technical First, we establish a technical point already alluded to in Section 3. The denition is
stated in terms of local martingales, rather than martingales, for the technical reason that the
former are easier to characterize in applications. Apart from a few embellishments necessitated
by some technical diculties, the ideas dier very little from those used to prove Lemma 4.
MATHEMATICAL ENGLISH USAGE 145
technicality This condition may seem unnatural, but it simplies some of the technicalities of
the proof. In order to bypass some technicalities, we also assume that..... Here the functional
analytic tools required are simpler, but this easier setting allows us to develop some methods
without undue worry about technicalities.
technique [see also: approach, method, procedure, way] His techniques work just as well for
general v. The technique is the same as in the proof of..... The argument is now completed
by means of techniques originated in the work of Stolz [3]. We encourage the reader unfamiliar
with techniques from the theory of..... to consult [BS]. A second technique for creating new
triangulations out of old ones is central retriangulation.
technology Although our proof is a little tedious, it is much less so than Itos original proof,
which was carried out without the benet of martingale technology.
tedious [= long and uninteresting] The proof is tedious but routine. Although our proof is
a little tedious, it is much less so than Itos original proof, which was carried out without the
benet of martingale technology. Kim announces that (by a tedious proof) the upper bound
can be reduced to 10. The only thing to check is that maps compose correctly; but this is an
easy, if tedious, verication.
tell [see also: say, assert, state, determine, decide] A direct application of Theorem 4 only tells
us that..... From the theory so far, we cannot tell if all four points are in the same Nielsen class.
To tell whether a labelled partial order is a skeleton, it suces to look at its substructures of
size at most three.
temporarily [see also: moment] Temporarily assuming that E is positive, we multiply.....
temporary [see also: provisional, ad hoc] Let us introduce the temporary notation Ff for gfg.
tempt It is also tempting to get round this problem by working with..... One is tempted to
reverse the order of integrations but that is illegitimate here. Having established (1), one might
be tempted to try to extend this result to general p through the choice of a suitable ideal B.
Alas, as we shall see now, this attempt is futile.
tend [see also: converge, approach] Clearly, F
n
tends to zero as n . Observe that as I
becomes smaller, d(I, f) tends to 1.
tentative [= cautious, not rm] However, a few tentative conclusions can be drawn.
term 1 [see also: name, notion, word, language] So all the terms of (2) are accounted for, and
the theorem is proved. This term drops out when f is dierentiated. Each of the terms
that make up G(t) is well dened. term-by-term dierentiation This theorem accounts for
the term subharmonic. The term upath is a mnemonic for unit speed path. The answer
depends on how broadly or narrowly the term matrix method is dened. The above denition
was rst given in [D]; care is required because this term has been used in a slightly dierent
sense elsewhere. The z-component can be expressed in terms of the gauge function. Our
aim here is to give some sort of functorial description of K in terms of G. This can be easily
reformulated in purely geometric terms.
2 [see also: call, designate, name] Garling has introduced a more general class of martingales,
termed Hardy martingales.
terminate [see also: end] a path terminating at x
146 JERZY TRZECIAK
terminology In order to state these conditions succinctly, we introduce the following terminology.
In Section 2 we set up notation and terminology. In discussing structures, we shall employ
the standard terminology of rst-order logic. This terminology is used slightly dierently in [6].
This is slightly at odds [= inconsistent] with the terminology of [4], as Fox denes the trace
lter to be the normal lter generated by A.
test [see also: examine, check, investigate] The property f(U) = V can be tested by examining
the behaviour of..... Computing f(y) can be done by enumerating A(y) and testing each element
for membership in C. We now test equation (6) with the function F to obtain.....
text Chapter 2 of the classic text [6] by R. Nevanlinna has a detailed treatment of this construc-
tion.
than Now F is greater than or equal to G. [Not: greater or equal to G, greater or equal G,
nor greater or equal than G] The point is that the operator is now much easier to analyse
than is the case in the original setting of the space B. Part of the conclusion is that F moves
each z closer to the origin than it was.
thank The author thanks H. Miller for a careful reading of an earlier draft. We thank the referee
for recommending various improvements in exposition. It is a pleasure to thank R. Greenberg
for bringing his criterion for..... to our attention, and for generously sharing his ideas about it.
thanks 1 Mary Lane deserves our special thanks for her part in bringing this volume to a success-
ful completion. The referee deserves thanks for careful reading and many useful comments.
We would like to express our thanks to all of these institutions for their hospitality and support.
2 Thanks to Lemma 2, we can now modify the proof of Theorem 3 by inductively constructing
ve sequences.....
that [see also: which, namely] That (2) implies (1) is contained in the proof of Theorem 1 of [4].
Clearly, A

weights are sharp weights. That there are no others is the main result of Section 2.
The degree of P equals that of Q. The continuity of f implies that of g. The diameter of
F is about twice that of G. It is this point of view which is close to that used in C

-algebras.
Dene f(z) to be that y for which..... Where there is a choice of several acceptable forms,
that form is selected which..... Associated with each Steiner system is its automorphism group,
that is, the set of all..... The usefulness and interest of this correspondence will of course be
enhanced if there is a way of returning from the transforms to the functions, that is to say, if
there is an inversion formula. We now state a result that will be of use later. A principal
ideal is one that is generated by a single element. Let I be the family of all subalgebras that
contain F. [Or: which contain F; you can use either that or which in dening clauses.]
the Note that the P produced in Theorem 2 need not have dP = 0. .....where the P
k
are
polynomials. We characterize the Banach spaces X for which n(X) = 1. Thus
n
(X) can be
interpreted as the homotopy classes of maps S
n
X. [= as the set of all homotopy classes] It
has some basic properties in common with another most important class of functions, namely,
the continuous ones. In the plane, the open sets are those which are unions of open circular
discs. Since u is constant on the level sets of a, it follows that..... the number of zeros of f
in D the density of the zeros of f Let A be the union of the sets f(Q) for f in F. Here p
2
(r)
is the sum of the squares of the divisors of r. The problem has a very natural connection with
the problem of the distribution of the zeros of a bounded holomorphic function in a half-plane.
MATHEMATICAL ENGLISH USAGE 147
Intuitively, entropy of a partition is a measure of its information contentthe larger the entropy,
the larger the information content.
theme [see also: subject, topic] We shall return to this central theme in Section 4.
then The complex case then follows from (a). Continuity then nishes o the argument.
Theorem 3 may be interpreted as saying that A = B, but it must then be remembered that.....
Then G has 10 normal subgroups and as many non-normal ones. If p = 0 then there are an
additional m arcs. [Note the article an.] If y is a solution, then ay also solves (3) for all a in B.
theorem Here is a more explicit statement of what the theorem asserts. What the theorem is
saying in substance is that..... This theorem accounts for the term subharmonic. Theorem 2
will form the basis for our subsequent results. [Not: The Theorem 2] In particular, the
theorem applies to weakly conuent maps. Finally, case (E) is completed by again invoking
Theorem 1. At this stage we appeal to Theorem 2 to deduce that..... Browns theorem
[without the] = the Brown theorem a theorem of Browns [= one of Browns theorems] .....,
which, by another theorem of Kimneys, is more than enough to guarantee that P gives A outer
measure 1. Wieners famous renowned/celebrated) theorem
theory In [2], this theorem is made the starting point of Gelfand theory. This approach is
standard in homotopy theory. Unfortunately, the notation from number theory slightly conicts
with the notation from probability theory. More specialized notions from Banach space theory
will be introduced as needed. This abstract theory is not in any way more dicult than the
special case of the real line. Kearnes developed a commutator theory for relative congruences,
with the expectation that it can be used to prove Pigozzis conjecture.
there Theorem 2, at the end of Section 2, was not originally obtained in the manner indicated
there. That approach was used earlier in [2]. There, however, it was applied in simply connected
regions only. This shows that there is r 0 such that..... [Or: there is an r 0] Now, there
are a and b such that..... There exists a function f and a constant c such that..... [Or: There
exist a function f and a constant c] However, there are a large number of examples showing
that..... [Not: There is a large number] If p = 0 then there are an additional m arcs. [Note
the article an.] We have to show that the property of there being x and y such that x < y
uniquely determines P up to isomorphism. Then (3.5) is a necessary and sucient condition
for there to be a function f such that..... How many such expressions are there? How many
entries are there in this section? In general, we have a b; there is equality if..... There has
recently been increasing interest in the theory of..... There is not space to enumerate them
all here. From the viewpoint of the Fox theorem, there is not an exact parallel between the
odds and the evens. For general linear operators, there is not such an extensive functional
calculus as there is for self-adjoint operators. Hence there can be no condition on the norms
which guarantees (7). In representation theory, there can never be a B-map whose domain
is nite-dimensional. There cannot be two edges between one pair of vertices. With each
D there is associated a region V
D
. There remain four intervals of length 1/4 each. There
remains one further case to consider. [But: It remains to consider one further case.] In [6]
there occur the following formulas. There seems to be no simple formula for .....
thereafter [= from that time on] This procedure, once implemented, can thereafter be applied
with great eectiveness.
148 JERZY TRZECIAK
thereby [= as a result of that] Clearly, A is thereby put in one-to-one correspondence with B.
So F must be constant on H, thereby showing that A = B, as desired.
therefore [= for that reason; see also: consequently, hence, thus] It is therefore of interest
to look at..... The polynomials R
i
do not have this stability property, and are therefore of
little interest. One must therefore also introduce the class of..... Therefore, whenever it is
convenient, we may assume that.....
these In this way, D yields operators D
+
and D

. These are formal adjoints of each other.

The best known of these is the Knaster continuum. One of these lies in the union of the other
two. We refer to these as homogeneous Sobolev spaces. Each of these three integrals is nite.
We begin by constructing a function f which has these prescribed zeros.
thesis [= dissertation] This was nally settled in Siebenmanns thesis. [Do not write: the
thesis of Theorem 3 if you mean: the assertion of Theorem 3.]
thing [see also: point, matter] Thus A-equivalence and B-equivalence are the same thing. The
only thing to check is that maps compose correctly; but this is an easy, if tedious, verication.
When r > 3 things become much more dicult. We begin by showing that -collections and
-collections are quite dierent things. The next corollary shows among other things that.....
[Not: among others]
think [see also: view, regard, consider] If one thinks of x, y as space variables and of z as time,
then..... It may help to think of F as being a smooth approximation to the Heaviside function.
This result may be thought of as a sort of regularity theorem. So s can be thought of as q
with F
q
extended but X
q
left the same.
third The budget has increased by more than a third. The algorithm examines only roughly
one-quarter to one-third of the characters. Each tree is about two-thirds as deep as it was
before. Let E be Cantors familiar middle thirds set.
this [see also: it] This is because the factor M satises..... This is why no truncation is required
here. (This is where the local convexity of E is needed.) This says that the real part of g
is..... We shall assume that this is the case. For K, this is no longer true. [Compare: For K,
it is no longer true that.....] In order to justify this, we now show that..... We leave this as an
exercise. In this and the other theorems of this section, the X
n
are any independent random
variables with a common distribution. In more detail, the assertion is this: if..... Adding over
n = 0, 1, ... leads to this result:.....
thorough [see also: extensive, complete, depth, detail] For a thorough treatment of..... we refer
the reader to Section 5 of [Ho].
those Those more than half a square count as whole ones. Using (2) and following steps
analogous to those above, we obtain..... Our results are closely related to those of Strang [5].
These intervals are disjoint from all those used in dening J
1
. In the plane, the open sets
are those which are unions of open circular discs. We dene the set T to consist of those f for
which..... Now A consists of those vectors with eight 2s. It is easy to see that the extreme
points of the unit ball M()
[1]
are those measures of the form
x
. Keep only those vertices
whose coordinates sum to 4.
though [see also: although, but, albeit] The sum in (2), though formally innite, is therefore
MATHEMATICAL ENGLISH USAGE 149
actually nite. For our application though, we need a stronger statement. Even though we
were able to derive a formula for....., it is not easy to use. The fact that the number T(p) is
uniquely dened, even though p is not, enables us to dene the nullity of A as follows.
three [see also: two] The three groups have the same number of generators. [= Those three
groups; not: The 3 groups; the numbers 1 to 12, when used for counting objects (without units
of measurement), should be written in words.] We now dene three more groups of interest.
We have thus found another three solutions of (5). [= three more]
through Note that T(g) depends on g only through its dierential dg. Then T tends to zero
only through positive values. Most distribution spaces can be characterized through the action
of appropriate convolution operators. Through an appeal to (5.3) we have..... Having estab-
lished (1), one might be tempted to try to extend this result to general p through the choice of a
suitable ideal B. Exercises 2 to 5 [= Exercises 25; amer. Exercises 2 through 5] We now go
through the clauses of Denition 3. On dividing through by f, we see that..... The rest of the
proof goes through as for Corollary 2, with hardly any changes. There are kneading sequences
for which the arguments of Section 4 go through routinely. Multiplying through by f and
integrating shows that..... This assumption enables us to push through the same arguments.
As t runs from 0 to 1, the point f(t) runs through the interval [a, b]. rotation through /3
through an angle )
throughout We adopt throughout the convention that compact spaces are Hausdor. Here and
throughout, P(E) denotes..... Throughout integration theory, one inevitably encounters .
Throughout what follows, we shall freely use without explicit mention the elementary fact
that..... The letter will be reserved for characteristic functions throughout this book. Then
a b c a. We therefore have equality throughout.
thus [see also: consequently, follow, hence, so, therefore] It will thus be sucient to..... [= there-
fore] Once this is done, the proof continues thus:..... [= in this way] The function thus dened
is a semigroup morphism. However, we have thus far been unable to nd any magic squares
with seven square entries.
tie [see also: connect, link, relate] The next theorem ties together the concepts of.....
tilde Generally we add a tilde to distinguish between quantities associated with

G and those
associated with G.
time [see also: moment, simultaneously] an impulse acting at time t = 0 This path stays in
B for all time. This can be performed in O(n) time. The problem is to move all the discs
to the third peg by moving only one at a time. At times [= Occasionally] it will be useful to
consider..... We prove both lemmas at the same time. At the time of writing [5], I was not
aware of this reference. It has been known for some time that..... But this time boundedness
on U is enough; we do not need continuity on V . Then A has three times as many elements
as B has. The integral of F is r times the sum of..... Thus X has r times the length of Y .
The diameter of A is three times that of B. Clearly, F is r times as long as G. Applying
this argument k more times, we obtain..... .....where z
k
= z...z (k times) Let K be the
number of times that z returns to B. Repeating this procedure enough times gives the desired
triangulation.
to Denote by the angle at x that is common to these triangles. The following lemma, crucial to
150 JERZY TRZECIAK
Theorem 2, is also implicit in [4]. Knowing this matrix is equivalent to knowing the multiplicities
of the l
i
. Thus will be less than by an amount comparable to a(s). Essential to the proof
are certain topological properties of G. We shall see later that the values of h(n) for large
n are irrelevant to the problem. We base our development on two properties of prolongation
peculiar to this case. But the T
n
need not be contractions in L
1
, which is the main obstruction
to applying standard arguments for densities. We write = ....., which is a slight modication
to the previous version of . We can join a to b by a path . [Not: join a with b] The case
where i is odd yields to a similar argument. Exercises 2 to 5 [= Exercises 25; amer. Exercises
2 through 5] To deal with (3), consider..... We now apply (2) to get Nf = 0. Theorems 1
and 2 combine to give the following result. To see that f = g is fairly easy. Every prime in
the factorization appears to an even power.
together [see also: combine, conjunction] Now (2) and (3) together are equivalent to (4).
Together with Lemma 4, this implies that..... It follows from this representation together
with Lemma 3 that..... ....., which, together with (2), shows..... The aim of this paper is to
bring together two areas in which..... These volumes bring together all of R. Bings published
mathematical papers. The key part is to show that the submanifolds U
k
t together to form
a complex submanifold. There is a natural way to glue the associated varieties together along
their common boundary. The next theorem ties together the concepts of.....
too [see also: also, well, similarly, likewise] In practice, D is usually too large a set to work with.
Consequently, A has two elements too many. [Or: A has two too many elements.] There
are other problems with this example which would hinder any attempt to follow the proof given
here too closely. We denote this, too, by Q. Note that this too is best possible. The inner
sum is zero (and so too is S(a, b)). If (1) and (2) hold, then so too does (3).
tool This is the essential tool in proving Theorem 5. These idempotents provide a useful tool
for analysing the structure of G. In this paper we oer some useful tools for bounding d-
pseudoprimes. The abstract theory gives us a tool of much wider applicability. A further tool
available is the following classical result of Chen.
top [see also: bottom, upper] at the top of page 4 near the top of the scale the top row
the top right-hand corner We now start piling the pieces on top of each other. In Figure 2,
the set A is marked by a square with a small triangle on top.
topic [see also: theme, subject, matter, point, issue] This topic has been dealt with by many
authors. Another topic of great interest is how much of adjunction theory holds for ample
vector bundles.
total 1 [see also: all, altogether] There are N vertices of the 24-cube which have a 1 in the ith
coordinate and a total of ve 1s. This makes a total of 170 elements adjacent to A. There
are twelve indecomposables in total.
2 [see also: whole, complete, entire, full] The total number of such vectors is 88. The total
amount of information lost is.....
3 [see also: add up, amount] The angles of a triangle total 180. Subsidies on these commodities
total 25 per cent of the budget.
touch [see also: refer, mention, allude] We shall touch only a few aspects of the theory.
towards [= toward] Suppose, towards a contradiction, that..... The vector eld H always
MATHEMATICAL ENGLISH USAGE 151
points towards the higher A-level. We now work toward the establishment of properties (A)
to (D).
tractable However, with the recent advent of simulation based inference, the need for analytically
tractable posteriors is no longer critical. Although individually these systems can still be quite
complicated, a possibly more tractable task is to describe their possible joint distributions.
transfer 1 [see also: passage, transition] Section 6 contains a formula which permits transfer of
the results in Section 2 to sums of independent random variables.
2 [see also: move] In this case it is advantageous to transfer the problem to (say) the upper
half-plane. We now transfer the above analysis back to M(A). Hence the action of G on A
transfers, via f, to an action on B.
transform [sth into sth; see also: alter, change, convert, make into, modify] Replacement of z
by 1/z transforms (3) into (4).
transition [see also: passage, transfer] Then we shall use (2.3) in order to make the transition
from M to a universal Turing machine U.
translate 1 It suces to take a suitable translate of U.
2 Section 2 then translates the results into equivariant cohomology. This can be translated
into the language of dierential forms.
transpire [see also: turn out] It transpires that F is not necessary here. [= It turns out]
treat [see also: consider, deal, discuss, present, handle] This subject is treated at length in
Section 2. In the present paper we move outside the random walk case and treat time-
inhomogeneous convolutions. The approach in [GT] provides a unied way of treating a wide
variety of seemingly disparate examples.
treatment [see also: consideration, discussion, study] For a thorough treatment of..... we refer
the reader to Section 5 of [Ho]. For a comprehensive treatment and for references to the
extensive literature on the subject one may refer to the book [M] by Markov. An extensive
treatment of the h-principle can be found in [6]. It is beyond the scope of this paper to give
a complete treatment of..... While the rst version was being refereed, I found that Zhang [2]
had given a similar treatment of E
n
(X). Chapter 2 of the classic text [6] by R. Nevanlinna has
a detailed treatment of this construction. The rst of these was suggested by J. Serrin, who
showed how to modify my earlier treatment of..... so as to obtain stronger results with no extra
eort. We could have adopted an approach to proving Theorem 2 along a line of reasoning
which bears greater resemblance to the treatment of the analogous result in Section 1.
trial [see also: attempt] A solution will be found by a process of trial and error.
trick [see also: subterfuge] In fact, there is a trick which enables one to reduce the time depen-
dent case on R
n
to the time independent case on R
n+1
.
trivial The trivial cases when the family consists of only one set are excepted. First we take up
the trivial case h = 0. If there are to be any nontrivial solutions x then any odd prime must
satisfy..... It is generally a highly nontrivial question whether.....
triviality To avoid trivialities, we shall assume that.....
152 JERZY TRZECIAK
trivially The equivalence of (a) and (b) is trivially implied by the denition of M. The estimate
is trivially true if A = B. The desired inequality holds trivially whenever A > 0.
trouble [see also: complication, diculty, problem] Now the reader will have no trouble verifying
that.....
true [see also: correct, valid, hold] The conjecture (now known not to be true in general) was
that..... In other algorithms, this may not be true. Unfortunately, because of the possible
presence of cusps, this need not be true. This is not true of F. [Or: for F] This works
regardless of whether B is true or false. Let us now take a quick look at the class N, with the
purpose of determining how much of Theorems 1 and 2 is true here. This suggests a question:
under what conditions is it true that.....? Only in exceptional circumstances is it true that
f(x +y) = f(x) +f(y). [Note the inversion.]
truncate We tabulate the outcome for n 10; in particular, the column headed R lists R(n)
truncated to three decimal places.
truth We will nish this section by oering a second application of our machinery (although in
truth it is largely a corollary of the above).
try [see also: attempt, eort, seek] In the remainder of this section we shall be trying to
answer the question:..... We shall try to give it the simplest representation possible. Having
established (1), one might be tempted to try to extend this result to general p through the choice
of a suitable ideal B.
turn 1 [see also: rotation] A turn through /3 restores the position of the needle. The
algorithm compares x with each entry in turn until a match is found or the list is exhausted.
The orbits of H on B are unions of orbits of N on G, which in turn are orbits of N on G
1
, G
2
and G
3
.
2 [see also: proceed, change into, make into, transform, appear, go back, transpire] We now
turn to a brief discussion of another concept which is relevant to Johns theorem. We now
turn to estimating Tf. Implementation is the task of turning an algorithm into a computer
program. We turn the set of..... into a category by dening the morphisms to be..... We now
turn back to our main question. It turns out that A is not merely symmetric, but actually
selfadjoint. This condition also turns out to be necessary. However, this equality turned out
to be a mere coincidence. Our main nding in this paper is that this intuition turns out to be
erroneous. It should come as no surprise that a condition like a
i
,= b
i
turns up in this theorem.
[= appears]
twice [see also: double] The curve C encircles the origin twice. The diameter of F is about
twice that of G. Let 2I denote the interval concentric with I but of twice its length. Any
point not in B is moved by f a distance equal to twice the distance to M.
two [see also: three] The player has to decide which of the two strategies is better for him and
act accordingly. [Not: the 2 strategies; the numbers 1 to 12, when used for counting objects
(without units of measurement), should be written in words.] The rst two are simpler than
the third. [Or: the third one; not: The rst two ones] One of these lies in the union of the
other two. the last two rows the following two maps a two-variable characterization
twofold Then P covers M twofold. The motivation for writing this paper was twofold.
MATHEMATICAL ENGLISH USAGE 153
two-thirds The sum of the depths is at most two-thirds of what it was before. Each tree is
about two-thirds as deep as it was before.
type [see also: kind, sort, form] A function exhibiting this type of behaviour has been con-
structed in [9]. It became clear that the Riemann integral should be replaced by some other
type of integral, better suited for dealing with limit processes. Section 2 presents two L
1
-type
characterizations. Then A is an algebra of type II. All of the action in creating S
i+1
takes
place in the individual cells of type 2 or 3. It is a simple matter to remove all type 1 edges.
Therefore, the system (5) has a solution of the sought-for type. for all f of the type specied
in Theorem 4
typical [of sth; see also: characteristic] This fact is typical of diraction.
typically A computational restraint is the algebraic number theory involved in nding these
ranks, which will typically be more demanding than in our example of Section 1. While
nonparametric priors are typically dicult to manipulate, we believe the contrary is true for
quantile pyramids. Here one typically takes E to lie in the subspace H.
U
ultimate [see also: eventual, nal] Our ultimate objective is to eliminate this assumption com-
pletely.
unaected [see also: unchanged] The right hand side of (3) is unaected if we replace H by G.
Properties involving topological centres are unaected by a change to an equivalent weight.
If a, b, and c are permuted cyclically, the left side of (2) is unaected.
unambiguously Formula (9) denes F(n) unambiguously for every n.
unaware [see also: know] In 1925 Franklin, unaware of Stackels work, showed..... There are,
however, a few important papers of which we were unaware until fairly recently. However, we
are unaware of any counterexample to the following conjecture:.....
unchanged [see also: unaected] The limit in 4(b) is unchanged if g is replaced by f.
unclear However, it is unclear how to prove Corollary 3 without the rank theorem.
under Under what conditions can f have a local minimum in A? Then X is a Banach algebra
under convolution multiplication under this norm). It follows that G is maximal under for)
the usual partial ordering of B. Hence F is invariant under . [= -invariant] the image of A
under f = the f-image of A The point x maps to under f. This class includes just under
3000 items.
undergo The shape of F undergoes radical changes as x moves from A to B.
underlie [see also: basis, foundation] It is this idea that underlies some of the results of [2].
The condition..... can be improved by employing a strategy similar to that underlying the proof
of Theorem 2. The semidirect product of H and G has H G as its underlying set. All our
results hold independently of whether the underlying eld is R or C.
underline [see also: emphasize, stress] We underline that the aforementioned results in [1] all
rely on the conformality of the underlying construction. This underlines the importance of the
Bass conjecture.
154 JERZY TRZECIAK
understand [see also: mean, know, realize, recognize, grasp] When we talk of a complex mea-
sure, it is understood that (E) is a complex number. The rst equality is understood to mean
that..... The ideal is dened by m =....., it being understood that..... To understand why,
let us remember that..... By quasi-equation we understand a sentence of the form..... Our
development will require a detailed understanding of cozero covers. Thus it is reasonable to
attempt, using this homeomorphism, to gain an understanding of the structure of M. It is
hoped that a deeper understanding of these residues will help establish new results about the
distribution of modular symbols.
undertake In [7] the authors undertook a detailed study of.....
undesirable However, if B were omitted in (1), the case n = 0 would imply Nf = 1, an
undesirable restriction.
unfortunately [see also: alas] Unfortunately, this is rarely the case. Unfortunately, the nota-
tion from number theory slightly conicts with the notation from probability theory.
uniform To handle the three cases in a uniform way, it is convenient to.....
uniformly Then F
n
(x, y) converges to F(x, y) uniformly in x. One must also be aware that
the curvature of M
i
might not be bounded uniformly in i. Hence, although the topology of
reducts of A is uniformly controlled, so to speak, by that of A, the model theory of the reducts
can be much wilder.
unify Very recently, Heck introduced a general approach that unies and extends all these results.
The approach in [GT] provides a unied way of treating a wide variety of seemingly disparate
examples.
unimportant These extra stipulations are unimportant, but are given for deniteness.
union Every open set is a union of balls. We have thus proved that B is the union of a collection
of balls.
unique [see also: single, only] To every F there corresponds a unique G such that..... This
equation has a unique solution for every p. Then F has a unique lift F

. Therefore, F has
the unique lift F

= p
1
F. Let F be the unique map such that..... Hence M is the unique
largest submodule of type (a).
uniquely This normalization determines V uniquely. Then G is uniquely determined up to
unitary equivalence. The fact that the number T(p) is uniquely dened, even though p is not,
enables us to dene the nullity of A as follows.
unit The m points x
1
, . . . , x
m
are regularly spaced t units apart. an interval of unit length
the unit mass concentrated at x an algebra with unit the units digit
unity The coecients of A add up to unity.
university After receiving his PhD he took a position at he came to) the University of Texas.
He received his master and PhD degrees from the University of Texas. In 1987 he went to
Delhi University. He is currently Professor of Mathematics at Texas State University. This
research was initiated when the rst author was visiting the University of Alberta in the summer
of 2008; the nancial support and kind hospitality are gratefully acknowledged.
MATHEMATICAL ENGLISH USAGE 155
unless We now prove that f cannot have compact support unless f = 0 a.e. We put b in R
unless a is already in. The S-function f is zero unless l = m, in which case it depends only
on M.
unlike [see also: contrast, oppose] Note that, unlike the maximal function, the Hilbert transform
is not..... Unlike Bells method, Halls does not use transnite induction. Unlike the case with
designs, there exist weak designs where the universe size does not depend on the number of sets.
unlikely This change is unlikely to aect the solution. We drop the subscript when confusion
is unlikely. It is unlikely that the disturbances will eventually disappear.
unnecessarily His proof is unnecessarily complicated.
unnecessary [see also: superuous, redundant] It is therefore unnecessary to specify G on M.
unresolved [see also: open] The question whether A belongs to W was left unresolved there.
until The algorithm compares x to each entry in turn until a match is found or the list is
exhausted. This process can be repeated until we obtain the promised triangulation. We
postpone the proof until Section 2. We walk from X to Y until the rst place where G
changes. We defer the proof of the moreover statement in Theorem 5 until after the proof of
the lemma. There are, however, a few important papers of which we were unaware until fairly
recently.
unusual [see also: remarkable, rare, strange] One unusual feature of the solution should be
pointed out.
up The saddle-point conditions are satised up to an error o(n). He used a new version of an
algorithm for nding all normal subgroups of up to a given index in a nitely presented group.
Here F is only dened up to an additive constant. Therefore, G is uniquely determined up
to unitary equivalence. If we know a covering space E of X then not only do we know that.....,
but we can also recover X (up to homeomorphism) as E/G. from stage A up to, but not
including, stage B The arguments from this point up to Theorem 2 do not depend on.....
Up to now, we have assumed that..... The coecients of A add up to unity. Other situations
in dynamics where the p-adic numbers come up are surveyed in [W]. Arguing as before, we
shall end up with a simple tree all of whose facets contain V . Neighbourhoods of points in
these spaces appear at rst glance to have a nice regular structure, but upon closer scrutiny, one
sees that many neighbourhoods contain collections of arcs hopelessly folded up. These subsets
join up to form a simple closed curve passing through A and B. The set WF(u) is made up
of bicharacteristic strips. Each of the terms that make up G(t) is well dened. Now F is
dened to make G and H match up at the left end of I. Write out the integers from 1 to n.
Pair up the rst and the last, the second and next to last, etc. Values computed for the right
side of (2) were rounded up in the fourth decimal place. In Section 2 we set up notation and
terminology. [= prepare] On TK we set up the symplectic structure induced by the metric.
[= introduce] This space of curves also shows up in the theorem of Meyer on..... We shall
split up K as follows. First we take up the trivial case h = 0. In closing this section we
take up a result which will play a pivotal role in the characterization of..... It should be no
surprise that a condition like a
i
,= b
i
turns up in this theorem. The lectures were written up
by M. Stong.
156 JERZY TRZECIAK
upon The second inequality follows upon considering R
i
for i > 0. We may assume, upon
replacing F by F
1
, that..... Upon combining the estimate for B with (5), we have now estab-
lished the rst conclusion of Theorem 8. Adding E to both sides of (1), we can call upon (2)
to obtain (3).
upper Kim announces that (by a tedious proof) the upper bound can be reduced to 10.
We use upper case letters to represent inverses of generators. [= capital letters] an upper
semicontinuous function
use 1 [see also: application, means, via, utility] Apart from these two chapters, we make no use
of the results of [4]. The proof makes essential use of the Sobolev inequalities. This section
makes heavy use of a theorem of Alsen and related results. Our proofs make substantial use of
classical topology of the plane. That isapart from the use of relaxed controlsprecisely the
stochastic Bellman equation. In the following applications use will be made of..... The idea
behind our use of the s is that..... We depart from these previous works in our use of the
non-ergodic versions of the basic machinery. We now state a result that will be of use later.
For later use in conjunction with the weighted averages occurring in (2), we next consider.....
[Note the double r in occurring.] For future use, choose any monotone h(m) tending to innity
such that..... The main new feature is the use of the face ring to produce lower bounds for
the number of vertices. The second way of constructing K(F) is through the use of..... Some
proofs may be simplied by the use by use) of.....
2 [see also: apply, employ, exploit] We shall not use this fact in any essential way. We obtain
(using the fact that Q is a probability measure) Q(A) =..... The advantage of using..... lies
in the fact that..... Quite a few of them are now widely used. We conclude with two simple
lemmas to be used mainly in the proof of..... This method for solving (1) was described in [5].
In [1] the methods used are those of dierential topology.
useful [see also: helpful, advantageous, valuable, convenient] A concept which has proved useful
in the study of measures is tameness. These results show that an analysis purely at the level of
functions cannot be useful for describing..... This assumption is useful for simplifying proofs.
The following variant of Theorem 2 is occasionally useful. This is the least useful of the four
theorems. At times [= Occasionally] it will be useful to consider..... It is useful to consider
some rather simple examples to gain some intuition. These idempotents provide a useful tool
for analysing the structure of G.
usefulness [see also: utility] The usefulness and interest of this correspondence will of course
be enhanced if there is a way of returning from the transforms to the functions, that is to say,
if there is an inversion formula.
usual [see also: common, customary, familiar, standard] As usual, we can rephrase the above
result as a uniqueness theorem. [Not: As usually] (with the usual modication for p = )
One unusual feature of the solution should be pointed out.
usually It will usually be assumed that..... The calculation of M(f) is usually no harder than
the calculation of N(f). In practice, D is usually too large a set to work with. This topology
is compact, but not usually Hausdor, nor even T
1
.
utility [see also: usefulness] The following result illustrates the utility of (3). These functions
are sometimes called elementary factors. Their utility depends on the fact that.....
MATHEMATICAL ENGLISH USAGE 157
utilize [see also: use, employ, apply] The middle part of Table 2 compares the classication
according to max a
i
, where only the longitudinal information is utilized, with those according
to max b
ik
, where both longitudinal and survival information are used.
V
vacuous In the case k = 0 condition (A) becomes vacuous.
vacuously If K is empty, part of the hypothesis is vacuously satised.
valid [see also: correct, true, hold] Theorems 3 and 6 of [2], with the appropriate changes, are
also valid. The statement of Theorem 5 remains valid if we replace f is compact by the norm
of f is bounded. However, as observed in [5], the proof given in [7] is equally valid for regular
sets. This argument is invalid for several reasons.
validate [see also: prove] Their remarkable achievement seemed to validate Johns claim. How-
ever, it turned out that.....
validity Altering nitely many terms of the sequence u
n
does not aect the validity of (9). All
the evidence points to the validity of the conjecture.
valuable [see also: useful, helpful, worth] Our study grew out of some valuable conversations
with Kirk Douglas.
value The iterates eventually reach the value 1. The map U(t) takes values in some compact
space G. Each factor in (4) has absolute value 1 on T. Let n
k
be the rst location to the right
of the kth decimal place of W that has a value less than b. Every element of A has f-value 2.
[= the value of f at this element is 2] Then F is less than 1 in absolute value. The terms of
the series (1) decrease in absolute value and their signs alternate. We claim that, by setting w
to zero on this interval, the value of F(w) is reduced. the largest k value
vanish [see also: disappear, zero] Hence F vanishes to order 3 to innite order) at x. a nowhere
vanishing vector eld
variable Thus X can be taken as coordinate variable on M. If one thinks of x, y as space
variables and of z as time, then..... a function of several variables [Not: of many variables]
Thorin discovered the complex-variable proof of Rieszs theorem. A homogeneous two-variable
function f can be reduced to a one-variable function F(x) = f(1, x).
variant [see also: adaptation, adjustment, modication, variation, version] The following variant
of Theorem 2 is occasionally useful. The argument is a variant of one in [5] and has been used
several times since.
variation [see also: adaptation, adjustment, modication, variant, version] The example is a
variation on the space known as the Warsaw circle. This is essentially a variation on the
prime number theorem. Let us sketch the proof of the rst estimate, which is a variation of
standard arguments. As noted in the introduction, this is basically combining Sawyers result
with a variation of the arguments of Hunt.
variety [see also: number, many, plentiful, several] Actual construction of..... may be accom-
plished in a variety of ways. The question of..... has been explored under a variety of conditions
on A. The approach in [GT] provides a unied way of treating a wide variety of seemingly
disparate examples.
158 JERZY TRZECIAK
various [see also: dierent, several] We shall be considering L on various function spaces.
We do not know how V depends on the various choices made. Few of various existing proofs
are constructive. The author thanks the referee for recommending various improvements in
exposition.
variously These spaces are dened and variously characterized in [1]. We should stress that
this is only one of several versions of the measurable selector theorem, due variously to von
Neumann, Jankow, Lusin and others.
vary [see also: alter, change, range, dier, dierent] Here c denotes a constant which can vary
from line to line. Fix n and let c vary. A natural question to ask is how the quantities A(S, T)
and B(S, T) vary as S and T change. As we let t vary, f(t) describes a curve in M. Within I,
the function f varies oscillates) by less than 1. The idea is that C is xed, but X and Y vary
according to circumstances. Then F varies smoothly in t. The samples vary in length. Note
that m is permitted to vary with the number of inputs. A number of authors have considered,
in varying degrees of generality, the problem of determining..... a slowly varying function
Computer evidence suggests the dynamics of these maps is rich and varied.
verication [see also: check, examination, inspection, proof, scrutiny] The only point that
requires care is the verication of..... A standard verication shows that..... This concludes
the verication of Claim 2. The only thing to check is that maps compose correctly; but this
is an easy, if tedious, verication.
verify [,= satisfy; see also: check, examine] We are seeking to verify the conjecture that every
Banach sequence algebra A that is a Ditkin algebra is PAA. To prove (8), it only remains to
verify that..... Indeed, it is routine to verify that the index so constructed is independent of the
choices made. We leave it to the reader to verify that..... ....., as is easily veried. Now the
reader will have no trouble verifying that..... [Do not write: The function F veries condition
(2) if you mean: The function f satises condition (2).]
version [see also: variant, variation] We should stress that this is only one of several versions of
the measurable selector theorem, due variously to von Neumann, Jankow, Lusin and others.
We will call on this version of the inverse theorem when we come to our applications in Section 2.
We write = ....., which is a slight modication to the previous version of . a current up-to-
date/preliminary/denitive/nal/amended/enhanced/plausible/rened/sophisticated/elaborate/
simplied/weakened/explicit/quantitative) version
very [see also: precise] These results therefore describe the very close connection between the
method of encoding and the structures we are aiming to classify. By its very denition, f is
continuous. Finally, (d) is clear from the very last statement of Theorem 4. A cycle may very
well be represented as a sum of paths that are not closed.
via [see also: means] The tangent space to N at x is identied with M via left translation.
Then F and G are homotopic via a homotopy H such that..... It is clear that A is related to
B via (3.4). The standard proofs proceed via the Cauchy formula. ....., from which it is an
easy step, via Lemma 1, to the conclusion that.....
view 1 [see also: viewpoint, idea, perspective] The preceding observation, when looked at from
a more general point of view, leads to..... It is this point of view which is close to that used
in C

-algebras. Both of these conditions are satised if f is bounded (the second in view of
MATHEMATICAL ENGLISH USAGE 159
Assumption 3). With a view to bounding I in (8) by the right side of (6), we rst.....
2 [see also: consider, regard, see, think] It is convenient to view G as a nilpotent group. The
theory of correspondences may be viewed as bridging the gap between..... Here we are viewing
the coecients as reduced fractions.
viewpoint [see also: standpoint, view, perspective] From the viewpoint of the Fox theorem,
there is not an exact parallel between the odds and the evens.
violate [see also: fail] We note that H is in fact not Lipschitz continuous if this condition is
violated. This violates the maximum principle.
visit 1 This work was conducted during a visit of the author to Dartmouth College.
2 This research was initiated when the rst author was visiting the University of Alberta in the
summer of 2008; the nancial support and kind hospitality are gratefully acknowledged.
visualize To calculate (2), it helps to visualize the S
n
as the successive positions in a random
walk. The knowledge of the invariant subspaces of an operator helps us to visualize its action.
W
want [see also: wish, desire, require] The reader might want to compare this remark with [2,
Cor. 3]. This module is denoted by H(X), or H(X, R) if we want to make explicit the coecient
ring. This t has the feature we want.
warrant [see also: merit, worth] A complete explication of the Fox spaces is warranted, as it
will likely reveal further clues to the dierences between the parabolic and hyperbolic theories.
way [see also: approach, method, procedure, technique, incidentally] Now G can be handled in
much the same way. Actual construction of..... may be accomplished in a variety of ways.
Thus modules over categories are in many ways like ordinary modules. We shall not use this
fact in any essential way. This abstract theory is not in any way more dicult than the special
case of the real line. An alternative way to analyze S is to note that..... Here is another way
of stating (c). However, we know of no way of deriving one theory directly from the other.
The approach in [GT] provides a unied way of treating a wide variety of seemingly disparate
examples. This is the way Theorem 3 was proved. This follows from Lemma 2 just the way
(a) follows from (b). It follows from the way f was dened that..... Any congruence arises
this way. Theorem 2 can be proved a number of dierent ways. Put this way, the question
is not precise enough. By way of illustration, here is an example of..... The implication one
way follows from Theorem 2. On the way we analyze the relationship between..... Along the
way, we come across some perhaps unexpected rigidity properties of familiar spaces.
weak The convergence of the sum on the left is of course a weaker statement than the convergence
of (2). In fact, we shall prove our result under the weaker hypothesis that W is weakly bounded,
rather than just bounded, on an innite subset of G. The weight satises a weak type (1, 1)
estimate. Then x
n
converges weak

to x.
weaken [see also: relax] The primary advance is to weaken the assumption that H is C
2
, used
by previous authors, to the natural condition that H is C
1
. We can make g Lipschitz at the
price of weakening condition (i). Note that (2) is a slightly weakened version of the Polya
inequality.
160 JERZY TRZECIAK
weakness [see also: disadvantage, shortcoming, fail] It is important to notice some of the
weaknesses inherent in the above approach.
wealth [see also: abundance, much] The survey article [5] by Diestel contains a wealth of
information about the Dunford-Pettis property.
well [see also: also, too] For binary strings, the algorithm does not do quite as well. But H
itself can equally well be a member of S. Our asymptotic results compare reasonably well with
the numerical results reported in [8]. Since the integrands vanish at 0, we may as well assume
that..... Other types t into this pattern as well. Note that both sides of the inequality may
well be innite. A cycle may very well be represented as a sum of paths that are not closed. It
may well be that no optimal time exists, as the following example shows. Although standard,
the notion of a virtual vector bundle is not particularly well known.
were Suppose the lemma were false. Then we could nd..... Suppose x were not in B. Then
there would be..... If it were true that....., the same argument would apply to f. However, if
B were omitted in (1), the case n = 0 would imply Nf = 1, an undesirable restriction.
what Under those conditions, what does the sum on the left hand side of (8) signify? Does
the limit of f(z) exist as z goes to zero? If so, what is it? What is F(c) if c is a positively
oriented circle? What relation exists between f and g? What about the case where q > 2?
By what has been proved, there exists n such that..... What is left is to show that..... What
is still lacking is an explicit description of ker C. The sequence a
n
is what is sometimes called
a recovery sequence for v. Here is a more explicit statement of what the theorem asserts.
The sum of the depths is at most two-thirds of what it was before. Throughout what follows,
we shall freely use without explicit mention the elementary fact that..... It is not immediately
obvious what this generalization has to be. We conclude that, no matter what the class of
b is, we have an upper bound on M. But if E is not reexive orwhat is the samew is
weak, then..... Here G is discontinuous and, what is more, it does not belong to V . [Note the
dierence between what and which in sentences similar to the last two examples: what refers to
what follows it, while which refers to what precedes it.]
whatever We shall, by convenient abuse of notation, generally denote it by x
t
whatever prob-
ability space it is dened on. The problem is that, whatever the choice of F, there is always
another function f such that.....
whatsoever The preceding denitions can of course equally well be made with any eld what-
soever in place of the complex eld.
when [see also: if, whenever] Then F becomes inner when extended to B. It is important to
pay attention to the ranges of the mappings involved when trying to dene..... When reading
the proof of Lemma 2, it is helpful to keep in mind that..... When is it the case that.....? We
have d(f, g) = 0 precisely when f = g a.e.
whence [= as a consequence of which] We have shown that....., whence it is readily inferred
that.....
whenever [see also: if, when] Whenever the dimension drops by 1, the rank drops by at most Z.
We denote the complement of A by A
c
whenever it is clear from the context with respect to
which larger set the complement is taken. Next, (1) shows that (2) holds whenever g = f(n)
MATHEMATICAL ENGLISH USAGE 161
for some n. Suppose that T

is continuous whenever T is. A semilattice A has breadth n i

whenever E < A and [E[ > n, there is an x such that.....
where Where there is a choice of several acceptable forms, that form is selected which..... Where
it is important to distinguish dierent norms on E, we shall use the notation..... Where it is
possible, we outline the proofs so that the reader will not have to hunt for obscure references.
Where we could, we have chosen these examples from naturally occurring mathematical struc-
tures. [Note the double r in occurring.] There are many situations where this occurs naturally.
This is where the notion of an upper gradient comes in. Important cases are where S =.....
[Note that where can sometimes be ambiguous: where i I after a formula can mean either
for all i I or for some i I.]
whereas [see also: while] Their result gives no information when k is large, whereas (5) is
signicant regardless of the size of k. Further, the argument here is in terms of A, whereas
that of Theorem 5 is in terms of the algebra B of Gelfand transforms.
wherever Also, wherever possible, we work with integer coecients, enabling us to obtain infor-
mation about torsion.
whether [see also: if] The question of whether B is ever strictly larger than A remains open.
[Or: The question whether] We shall be interested in seeing whether..... Note that this lemma
does not give a simple criterion for deciding whether a given topology is indeed of the form T
f
.
So far it seems not to be known whether the geometric condition on X can be omitted. We do
not know whether or not Q(R) = R in this situation. To tell whether a labelled partial order
is a skeleton, it suces to look at its substructures of size at most three. Perhaps the most
important problem involving f-vectors is whether or not Johns conditions extend to spheres.
We conclude that whether a space X is an RG-space is not determined by the ring structure
of C(X). The method works irrespective of whether A or B is used. Unfortunately, the details
of the calculations were omitted, and there is some doubt on whether the result is correct since
our analysis suggests that P
2
must vanish to third order; the presence of L
2
is also suspect.
Thus, whether or not x is in the list, one comparison is done. The derivation D is the same
whether we regard E as a derivation on X or on Y . We consider every subset of N, whether
nite or innite, to be an increasing sequence.
which [see also: that] For which f does equality hold in this inequality? There has since been
a series of improvements, of which we briey mention the work of Levinson. Where there is
a choice of several acceptable forms, that form is selected which..... Let I be the family of all
subalgebras which contain F. [Or: that contain F; you can use either that or which in dening
clauses.] We denote the complement of A by A
c
whenever it is clear from the context with
respect to which larger set the complement is taken. The map f, which we know to be bounded,
is also right-continuous. [Not: that we know; do not use that in non-dening clauses.] Hence
F is compact, which yields M = N. [Not: what yields] ....., which completes the proof.
[Note the dierence between what and which in sentences similar to the last two examples: what
refers to what follows it, while which refers to what precedes it.]
whichever It produces the same outcome whichever path is taken. [= no matter which]
while [see also: whereas] In 1988, while attempting to generalize this result, the second author
noticed that..... While topological measures resemble Borel measures, they in general need not
be subadditive.
162 JERZY TRZECIAK
whole [see also: all, complete, entire, full, total, altogether] We can extend f by zero to the
whole . [Or: the whole of ] Those more than half a square count as whole ones. On the
whole, the solution can hardly be considered satisfactory. a fragment of a greater whole
wholly [see also: completely, entirely, fully] This book is wholly concerned with.....
whose Thus C lies on no segment both of whose endpoints lie in K. a manifold all of whose
geodesics are closed [= a manifold whose geodesics are all closed] a progression each of whose
terms can be written as..... Let M be the manifold to whose boundary f maps K. There
are n continua in X the union of whose images under f is K. Here x is the set whose only
member is x. Let A
i
be disjoint members of M whose union is E.
why To understand why, let us remember that..... This is why no truncation is required here.
Now why can such objects be found?
wide [see also: large, broad, extensive, substantial] We give a fairly simple description of a
wide class of averaging operators for which this rate of growth can be seen to be necessary.
The approach in [GT] provides a unied way of treating a wide variety of seemingly disparate
examples. The abstract theory gives us a tool of much wider applicability. This class is wide
enough to include a number of examples of interest.
widely Quite a few of them are now widely used.
-wise Let A denote the rectangle B rotated through /6 in a clockwise direction about the vertex
(0, 1). Pointwise convergence presents a more delicate problem. The sequence f
n
converges
to f pointwise. [Not: pointwisely] The set S is a semigroup with respect to coordinatewise
addition.
wish [see also: want, desire] We wish to arrange that f be as smooth as possible.
with The terms with n > N add up to less than 2. an algebra with unit a function with
compact support Then F is Poisson distributed with mean m. With this denition, the set of
equivalence classes is a metric space. With the customary abuse of notation, the same symbol
is used for..... It is proved in [1] (albeit with a slightly dierent formulation) that..... With
more work, one could show that..... .....with e
0
denoting multiplication by f. As with the
digit sums, we can use alternating digit sums to prove..... [= Just as in the case of digit sums]
Choose in accordance with Section 8. In analogy with (1) we have..... With each D there is
associated a region V
D
. We show that A is negligible compared with B. This notion is closely
connected with that of packing dimension. Let us continue with the proof of Theorem 2. It
is worth making a link with Theorem 1. The exact sequence ends on the right with H(X).
a word starting with a and ending with b We can make this clear with the following example.
Note that m is permitted to vary with the number of inputs.
within [see also: inside, up to] The point p is within distance d within a distance d) of X.
Then F is within d of the integers. The zeros of L-functions are all accurate to within 10
5
.
Note that f is determined only to within a set of measure zero. Each component that meets
X lies entirely within X. Within I, the function f varies by less than 1. This example falls
within the scope of Coxs theorem. a sequence of smooth domains that approximates D from
within
without This shows that f could not have n zeros without being identically zero. Hence Z enters
MATHEMATICAL ENGLISH USAGE 163
D without meeting x = 0. Thus C can be removed without changing the union. This allows
proving the representation formula without having to integrate over X. Take g
1
, ..., g
n
without
common zero. However, it is unclear how to prove Corollary 3 without the rank theorem.
Throughout what follows, we shall freely use without explicit mention the elementary fact
that.....
witness 1 Take for T any witness to X having property P.
2 [see also: exemplify] Now T witnesses property P. Note that R
2
witnesses that A is compact.
wonder The reader may wonder why we have apparently ignored the possibility of obtaining a
better lower bound by considering..... It is natural to wonder whether the inequality remains
true for the more general situation in Theorem A.
word [see also: name, term] We add the word positive for emphasis. The words collection,
family and class will be used synonymously with set. Let us state once more, in dierent words,
what the preceding result says if p = 1.
work 1 The present paper owes a great debt to Strangs work. Recent work of Kirby shows
that..... This work was conducted during a visit of the author to Dartmouth College. Kirk,
building on work of Penot, developed a more abstract version of..... The name of Harald Bohr
is attached to bG in recognition of his work on almost periodic functions. With a little more
work we can prove.....
2 [see also: succeed] In addition to illustrating how our formulas work in practice, it provides
a counterexample to Browns conjecture. The method works regardless of whether A or B is
used. The argument works equally well for R being a general ring with unity. A further
complication arises from BP, which works rather dierently from the other labels. We can
assume that p is as close to q as is necessary for the following proof to work. The denition of
generator is designed to make the proof above work for M = Z. Also, wherever possible, we
work with integer coecients, enabling us to obtain information about torsion. Rather than
working directly with V (s), we shall instead consider the following two general integrals:..... [Or:
Rather than work] In practice, D is usually too large a set to work with. We now work toward
the establishment of properties (A) to (D). In Section 4 we worked out a fairly detailed picture
of linear H-systems.
worth [see also: merit, warrant, worthy] It is an idea worth carrying out. [Not: worth while
carrying out] It is also clear that there are extensions to the case of....., but they do not seem
to be worth the eort of formulating them separately. Before going to the proof, it is worth
noting that..... It is worth making a link with Theorem 1. It may be worth reminding the
reader that.....
worthy [see also: merit, worth, warrant] The method sketched in Section 3 of [Con] carries
through with our choice of =
1
+
2
, but there is one extra ingredient worthy of mention.
would Hence we would expect the functions f
j
to behave similarly. Compact multipliers, as
one would expect, are those elements of A which..... One might hope that this method would
work at least for suciently regular maps; however,..... It would clearly have been sucient
to assume..... If it were true that....., the same argument would apply to f and would show.....
However, if B were omitted in (1), the case n = 0 would imply Nf = 1, an undesirable
restriction.
164 JERZY TRZECIAK
write [see also: denote, symbol] We write H for the value of G at zero. We shall frequently
write w.w. for weakly wandering. ....., which we can write as Df =..... Such cycles are
said to be homologous (written c c

). To simplify the writing, we take a = 0 and omit the

subscripts a. At the time of writing [5], I was not aware of this reference. In the course of
writing this paper we learned that P. Fox has simultaneously obtained results similar to ours in
certain respects. Important analytic dierences appear when one writes down precisely what
is meant by..... It may be dicult to write down an explicit domain of F. Note that (2) is
simply (1) written out in detail. By writing out the appropriate equations, we see that this is
equivalent to..... The lectures were written up by M. Stong.
Y
year [see also: date] The same year, Maurer showed that..... Maurer, in the same year, showed
that..... In the year 2000 In 2000), two important number theory conferences were held at
Princeton University. However, the connection with Gromovs work has been obscured in recent
years by an emphasis (in the algebraic topology literature) on conguration spaces.
yet [see also: but, still, nevertheless] For j = 1 the operator is bounded, yet [= but] the integral
(8) fails to be nite. Then F is strictly increasing and yet has zero derivative on a dense set.
In the next section we introduce yet another formulation of the problem. The as yet unproved
conjecture of Newman is that..... On the other hand, as yet, we have not taken advantage of
the basic property enjoyed by S: it is a simplex. The conjecture has not been proved yet.
In the remainder of this section, we study some properties of K, with the eventual aim (not
realized yet) of describing K directly using G. Further research may yet explain the enigma.
yield [see also: produce, provide] The argument can easily be modied to yield a proof for the
case of k positive. At rst glance Lemma 2 seems to yield four possible outcomes. Another
proof (yielding more information) can be found in [GH]. Specically, one might hope that a
clever application of something like Choquets theorem would yield the desired conclusion. The
idea of the following proof, which yields both (a) and (b) at one stroke, is due to von Neumann.
This nally yields f = g. [Not: yields that f = g] The case where i is odd yields to a
similar argument.
Z
zero [pl. zeros or zeroes; see also: vanish] Equating the coecient of x
2
in V to zero, we get.....
Then F has simple zeros with residue 1 at the integers. The integrand is zero outside D.
the multi-index with all entries zero except the kth which is one the zero solution the all-zero
vector