Robert David Sack

A Concept of Physical Space in Geography

The use of physical space in the laws of science, especially physics, has been
the focal point of discourse in philosophies of space and time for centuries.
Most of the current philosophical thought about space (and time) Concerns
physical theories of the cosmos which have as their subject the matter and
energy of the universe. The more circumscribed sdences, among them the social
sciences and geography (both physical and human), have also assumed physical
space to be a useful concept in their laws and theories. In these sciences, how-
ever, the concept of physical space has not been given the attention it has in
physics. In spite of the omnipresence of “spatial” terms in geography and their
occurrence in the social sciences, much of what has been said about them, and
many of the explanations that employ them, are imprecise because these
disciplines have been unclear about the meaning of the terms “physical space”
or “properties of physical space.’’ In this paper, we define “geographic” space
as physical space and confme our discussion to the concept of physical space
and its meaning in geography.
Most geographers and physicists contend that the three dimensional Euclidian
space with meters or miles as its units is the geometric system that adequately
describes the structure of the space of the physical world. The selection of the
geometry of classical physics by geographers as the structure of the space as-
sociated with the occurrence of geographic events is not a controversial point.
The theory of relativity confirms the appropriateness of Euclidian geometry for
terrestrial events and for most phenomena in our solar system.l What is at issue

1To daim that geographic space should refer to several or all synthetic spaces is to make
the term geographic space practically meaningless, espeaally since some synthetic functions
may be seen as “mappings” of one “space” into another, where the properties of either may be
metrical. (For a definition of synthetic see Carnap [Sl). Therefore mention in the geographic
literature of multi-dimensional spaces and non-Euclidian spaces, such as Bunge’s discussion
of “hyper space” [7, p. 2721, Adam’s et al. reference to “relative” space [ I ] and Harvey’s
discussion of distances [ l o , p. 2101 will be considered as references to non-geographic syn-
thetic spaces.

Robert Sack is assistant professor of geography at the University of Wisconsin,

Madison, Wisconsin.
 Robert David Sack / 17

is that in spite of the consensus about the geometric structure of physical space,
there appears to be confusion in geography (and in other social sciences) about
its meaning, an issue which has effected the empirical as well as the philosophical
basis of the discipline.
In geography, instances of the space of the physical world are often expressed
as events in themselves, independent of their connection to substance or matter,
and these expressions are expected to be significant concepts in geographic
hypotheses or laws. It is our contention that the expression of physical geometric
terms alone, unmodified, can be used without ambiguity, only in the laws of the
geometry of the physical world (or the laws of physical geometry, or physical
geometric laws). But these are not the laws of geography, nor are they all the
laws of science. In order for the concept of physical space to be useful in
geographic hypotheses (as well as in the hypotheses of other sciences) it should
conform to the criteria of concept formation in science. These criteria appear
to introduce factors external to the description of physical space, factors which
modify spatial terms.
According to many philosophers of science, the criteria that should be met
if a concept is to be useful in science, are that (1) we must be able to identify
one or more instances of the concept (that it have referents), and (2) we must
be able to determine if the concept enters into laws with other concepts in the
discipline (that it be significant) [2, 5, 61. The “referent” of the term can be
called its meaningl, and the “significance” of the term can be called its mean-
ingz. Together the referent of the term and the significance of the term supply
a concept with a meaning [ 6 ] .Physical space should have both of these meanings
if it is to be a useful concept in science. In this essay the expression “meaning
of physical space” encompasses both meaning1 and meaningz.
I contend that for a concept of physical space to conform to the rules of
concept formation and be useful in a science of geography every instance of the
geometric or spatial terms must be connected or related to one or more instances
of non-geometric terms (to be called substance terms). The connection involves
naming within the context of the hypothesis or law, the significant substance
referents of the geometric terms. When such a connection is made, we will call
the statement a “relational concept of space.”a Such a concept of physical space
appears to be the only one acceptable as a scientific concept, yet it has not been
extensively employed in the discipline.8 Instead, the frequent expression of
physical spatial terms without named substance referents in geography suggests
either a concept more akin to an absolute space, which denies that the criteria
of meaning are necessary for a useful spatial concept, or a poorly conceived
2 The expression “relational concept of space” is used here to refer only to the connection
between physical geometric terms and substance terms and to the requirement that the sub-
stance referents of physical geometric terms be made clear in non-geometric laws or hypothe-
ses. It is a more spedfic statement than the “relational theory of space” [24, pp. 108-1161
which appears in several cosmologies.
8Blaut [4and May [231, among others, have suggested that substances have geometric
properties and thus all sciences are examining “spatial” questions. This position is in perfect
accord with the relational concept of physical space, but it is not itself the relational position,
for it does not necessarily assert the converse, that every instance of a physical spatial prop-
erty is associated with instances of substance terms and that the signiscant substance referents
of geometric terms must be made clear in non-geometric laws or hypotheses.
18 / Geographical Analysis

relational one, which lacks complete or precise substance referents. Both con-
cepts result in similar and inadequate statements of physical space, hindering
the development of testable propositions and our understanding of geography’s
place among the sciences.
The paper is divided into four sections. In the first, I discuss some examples
of the assertion in geography that properties of physical space without named
substance referents are useful concepts in other than geometric laws. I n the
second section the inadequacies of justifications for this assertion are considered.
I n the third section I introduce the basis for the relational concept of space by
discussing the manner in which instances of physical spatial terms are as-
sociated with instances of substance terms as a means of identification. The
fourth section, by considering the use of spatial terms in scientific laws, ex-
plicates the relational concept of physical space and its implications for geo-
graphic hypotheses.

I. USE OF A NONRELATIONAL
CONCEPT OF PHYSICAL SPACE

The use of physical spatial terms without named substance referents occurs
in a variety of locutions throughout the geographic literature. As an illustration
of the frequency of their occurrence, it is sufficient here to remind ourselves of
the ubiquity of such expressions as “the use of physical space,” “the effects of
physical space,” “the effects of distance and location on behavior,” “the effects
of form, shape, or morphology on function.” Our concern in this paper, however,
is with the use of physical spatial terms in those statements of geography which
are offered as “scientific” hypotheses, ones that are supposed to be precise and
testable. It is in such hypotheses that we also find a profusion of spatial terms
without named substance referents. Often the use of “unmodified” spatial terms
is justified by declaring the terms to be “surrogates” for significant, though
unspecified, substances. It is our contention, however, that unless these “sur-
rogates” have named substance referents, these hypotheses, as stated, are un-
confirmable propositions. If some assert that they were never “scientific” in the
sense of being precise and testable to begin with, then perhaps our analysis can
help explain why.
Among those geographic statements which can be considered subject to scien-
tific scrutiny, the most general and persistent assertion of the significance of phys-
ical spatial terms without named substance referents is found in the “distance
interactance” or “distance decay” hypothesis. The hypothesis may be divided into
two parts [ 2 0 ] . (1) It asserts that ceteris paribus, interaction will occur more
frequently and/or with greater intensity, more facility or less cost, with places
near than with places far. (2) It asserts that the relevant measure of near and
far is the distance function of the space in which the events are located: namely,
the three-dimensional Euclidian space with miles or meters as its units. The
first part specifies that distance is significant because it is connected to inter-
action. If we were not to specify what we meant by distance and how it was
related to events, but to insist instead that if we select the proper function and
units of distance, a relationship between distance and events might emerge, we
 Robert David Sack / 19

would then be declaring only that we should be looking for useful measures.
The second part of the hypothesis appears to avoid such a vacuous statement
by specifying that the relevant measure of near and far is defined by the distance
function of the physical space. Thus instances of physical space, expressed as
distances, are assumed to occur and are expected to be significant.
What though are instances of the hypothesis? A minor problem in verifying
the hypothesis is that interaction is not clearly defined. To overcome this dif-
ficulty, numerous clear and precise definitions of interaction have been offered,
from the volume of telephone calls, to shopping behavior, to the frequency of
marriages, to the spread of information; and in several empirical studies
negative sloping curves exist which relate one or several of these clearly specified
measures of interaction to physical d i~ t a n c e .~
But it has also been found that
the correspondence between a specific interaction and physical distance may
change from place to place even within a culture, to the point where the slopes
of the curves change drastically, in some cases producing bulges or discontinu-
ities. Furthermore, there are many examples of well-defined instances of human
behavior where the selection of alternatives may correspond negatively or not
at all to physical distance.
The major problem with verifying the hypothesis does not lie in the vagueness
of the interaction term but rests instead with the use of physical distance with-
out specific substance referents. There are an infinite number of physical dis-
tances (even of a specified length) throughout the physical world. But the
significance of these distances varies depending on the substances of which they
are lengths. Cities, villages, and people are separated by lengths of things like
roads, paths, telephone wires, and air. Unless these things are specified we will
not know which lengths are or are not significant. The empirical works from
which the data on interactions are derived involved the examination of spatial
properties of things (space is not empty). But the introductions and conclusions
of “spatial” interaction studies discuss correspondences between specific inter-
actions and physical distance, where the significant substance referents of the
spatial terms are frequently omitted in the generalizations. Although these
studies are accurate descriptions, most cannot be generalized, for to make them
general we must know the significant substance referents of the geometric terms
to tell if the same conditions apply elsewhere, or a t another time. Thus, even
when interaction is clearly specified in the hypothesis, because distance remains
“unmodified,” the hypothesis remains unconfirmed. What illustrations there are
of precisely defined interaction decreasing with “unmodified” physical distance
are in Rushton’s terms descriptions of “behavior in space,” rather than postu-
lates of “spatial behavior” [ 18,p. 3921.
Among the more elaborate geographic hypotheses, perhaps the clearest as-
sertion of the significance of properties of physical space alone, independent

4 The number of empirical studies relating distance to interaction is enormous. Olsson’s

work on Distance and Human Interaction [I61 is an excellent review of the literature up to
the middle 1960s. I know of no comparable review for more recent material. Morrill and
Pitts’ work on “Marriage, Migration and the Mean Information Feld” 1141 illustrates
well some of the “behavioral” studies of distance and interaction that were not foreshadowed
in Olsson’s review. For a critique of the distance decay hypothesis see Rushton [IS].
20 / Geographical Analysis

of substance, is found in the statements by Warntz and Stewart on social physics.

Although these formulations have for long been questioned [I21 they are still
excellent illustrations of the difficulty that is encountered when unmodified
spatial concepts are used in “scientific” hypotheses.
In his discussion of the function of physical space in social physics, Warntz
declared:

... space and time are to be recognized not just as incurring external frictions,
but rather as dimensions of the economic system and hence to be treated iso-
morphically in the rigid pattern of mathematical physics. [ZS, p. 128.1

And Stewart stated:

In any physical situation alteration of the power [i.e., the exponent of one for
the distance term] would be a serious matter, not one merely of the choice of an
. .
adjustable parameter. . The “weight” assigned to the people is not similarly
critical; it can and must be adjusted to fit the observations. [Zl,p. 153.1

The exigencies of empirical observations, however, made adjustments in the

exponents of “space” and “time” as plausible as the alterations in the “units”
and “weights” of the masses. The consequences have been changes in the
models to make the distance variable flexible, all of which is exemplified by a
general expression of the gravity model such as

in which Zt, is the volume of interactions between places i and j , W, and w, are
empirically determined weights, Pc and Pj are the population sizes of places i and
j , Dlj is the distance between places i and j , and b is an exponent measuring the
friction of distance [27, p. 1141.
Instead of attempting to relate each adjustment of the “d” term to the vari-
ations in substances involved in transportation, and thus have terms referring to
distances of substances (i.e., lengths of railroads, rather than lengths), there
appears to be the expectation that physical distance can be salvaged as a
variable by itself [ 121.
We do not deny that lengths and other physical geometric properties occur
and that they can be given clear and unambiguous meanings (ml), but main-
tain that for anything but the laws of physical geometry, a spatial term that
does not refer to a specified substance is insufficiently defined for it to be a
significant concept. Suppose for example that the straight line (perhaps more
accurately the geodesic) separating city “a” from city “b” is the same length
as the line separating city “c” from “d.” The line between “a” and “b” however
“goes over” water, while the line of the same length between “c” and “d” is
“over” desert. Do they tell us the same thing? Will either be significant in the
gravity or potential model? How can we assert more than that they tell us that
the lengths between the two cities are the same? To determine if a length is
 Robert David Sack / 21

significant in any other way than geometrically, knowing the length must tell
us the substances which compose it. And even then, the length of the substance
may not be a significant concept in a particular hypothesis. Perhaps, one may
argue, it is not the straight lines at all that we are interested in, but rather the
“paths” or “routes” that will be taken between the two sets of places. But paths
are lengths of substances and it is an empirical matter whether a length (or any
other geometric property) of a particular substance or substances is a significant
concept in a particular hypothesis.
The expression of spatial terms, independent of substances, however occurs
in all other areas of geographic literature. It is at the base of central place theory
in the form of range, of Von Thiinen’s theory in the form of distance to the
market, in urban land use theories in the form of distance from the central
business district, and forms part of the concepts of “action spaces,” “awareness
spaces,” and “mean information fields.” With the assumption of isotropy in the
classical location theories, the connection between geometric properties and
substances is explicitly severed. Often the arguments for isotropy are put in
terms of uniform cost surfaces, where the coordinates are such things as dollars
or time. If, however, we want to connect these surfaces to coordinates of physical
space, the significant substances which compose these transport systems, and
give rise to costs, must be known to determine if the costs vary as a function
of their geometric properties. Even if these substances were known, their
geometric properties may in no way be regular and repetitive over the surface
of the earth.
The conspicuousness of the assumption of the significance of spatial properties
independent of substances is reflected in Bunge’s statement that the central
problem of geography “is the problem of placing interacting objects as near to
each other as possible, or simply, the nearness problem” [7, p. 331. When
observation has shown physical spatial properties not to be significant, the
behavior is often deemed irrational or non-geographi~.~ But the usefulness of
a concept in science is not a normative but an empirical matter. Empirically
we may determine the significance of the geometric properties of specific sub-
stance in non-geometric laws, but not simply the significance of geometric
properties.

11. “JUSTIFICATIONS” OF NONRELATIONAL

CONCEPT OF PHYSICAL SPACE

In their attempts to justify assumptions that properties of physical space are

significant concepts in geographic hypotheses, geographers assume that instances
of physical space can be expressed without connections to instances of sub-
stances. In this regard, an examination of these “justifications” may further
illustrate geography’s confusion about the meaning of physical space. Perhaps
the most frequently offered “justifications” are the “intervening opportunity
model” and the assumption of the “frictional effects of distance” (or simply

6See for example Morrill’s use of the terms “spatial error” or “spatial inefficiency” [15,
pp. 175-2021 and Bunge’s use of the term “inefficient movement” C7, p. 2001.
 22 / Geographical Analysis

the “friction of distance”)? A third “justification” will be presented, which,

although not stated explicitly in the geographic literature, is based on the
tautological properties of Euclidian geometry and may further illustrate our
point.
The essence of the intervening opportunity model can be stated briefly as the
assumption that, provided equal substitutability of places exists within a set of
places considered, interaction will be inversely proportional to the cumulated
points (substitutable places) encountered along a sequence. Clearly, the model
is concerned with the ordinal measure of places in physical space. But what is
the sequence or order of encounter? To define the sequence as the order of
equally substitutable places found lying along a straight line in physical space
is permissible, but would forfeit the model as a justification of the usefulness
of physical space, for it is precisely the question of why and when such an
order of events would be relevant that is at issue. It can be assumed that the
arrangement of places in physical space does not determine the order of en-
counter. However, if this is the case, then knowledge of the location of these
places in the geographic space would not be useful. Furthermore, unless we knew
the process by which places were encountered, the model would amount to the
weak declaration that if the sequence of encounters were known, its ordinal
properties would be useful. This however is not a justification of the significance
of properties of physical space but rather an expression of it.
The phrase “friction of distance” (or “frictional effect of distance”) is some-
times used as a synonym for distance decay. If we assume that what is meant
by i t is a reference to the measure of friction in physics then we could say that
friction of distance refers to the cost in terms of work done in overcoming fric-
tional resistance of relative motion. The issues here are (1) how is the distance
measured and related to the cost and ( 2 ) is the measure of cost useful in
geography? As to the first issue, friction in physics is due to the properties of
the phenomena involved. Work must be done, or energy expended or dissipated,
to overcome friction. The connection of the energy expended to an appropriate
measure of distance depends on the medium and on whether the force is con-
servative or not. If i t is, then the work done, although a function of the medium,
would be path independent. If the force is nonconservative, then the work done
would depend on the path followed as well as on the nature of the substances.
These considerations, however, are dependent on the substances involved and
are too complex to be called a “friction of distance,” and in fact, in physics
there is no such concept. There are frictions and there are distances, but there
is not friction of distance. Unless we know the phenomena we will have no idea
what the expenditure of energy in physical units would be, nor how it would
be related to physical distance. What is more, friction and energy are concepts
of physics, which brings us to our second consideration.
The usefulness of these physical concepts for geography is purely an empirical
matter, as is the usefulness of any concept, and depends on whether instances
of them are related to instances of other concepts. To claim that, because friction

6The authorship of the “intervening opportunity model” rests with Stouffer [223. I am
unaware of the originator of the expressions “frictional effects of distance” or “friction of
distance.”
 Robert David Sack / 23

occurs and that to overcome it requires expenditures of energy which may in-
crease as some measure of distance, does not mean that this physical cost is
significant in geography’s explanations, or that it is a justification of why
physical distance should be significant. The costs incurred in overcoming
physical resistance or friction may be insignificant for many processes, especially
human ones, and may have nothing to do with determining the amount of
interaction between places or the cost or facility involved in moving. If friction
were categorically useful in geography, then coefficients of friction (not friction
of distance), would be included in geographic explanations. If it were argued
that by friction of distance was meant the cost that would be necessary to
overcome physical separation and that this cost was a function of distance and
was useful in geography’s explanations, then we would no longer be talking about
friction in physics, but rather of the distance decay hypothesis in geography.
Since properties of physical space are asserted to be useful, then perhaps it
might not be unreasonable to expect that the justification lies in the nature of
Euclidian geometry itself. The tautological structure of the analytic Euclidian
geometry does, under certain conditions, provide us with what may be termed
a “decay function.” However, the difficulty in interpreting these results is
extremely significant and its consideration may further illustrate the problems
involved in expressing space independently of substances.
Consider a Euclidian space in which all events or phenomena are absent. The
only means by which one place can be differentiated from another is by its
location relative to an origin (that is, by its distance and direction). Suppose
this Euclidian space had three dimensions and we selected a single point in the
space and labeled it p. From this point we can imagine a finite, though large,
number of lines radiating out evenly in all directions, undiminished, until the
end of space. Now if we were to assume that p was the center of a sphere, SPO
having radius YO, the surface area of the sphere would be SO= 4mO2. Every line
emanating from p intersects the surface of sphere SPo. If YO were increased to
r1, r2,. .. , Ym) while still retaining p as the center, then there would be n
corresponding concentric spheres SPX, SP2, ... , SP,,,each having surface
areas equal to 4mm2.All of the lines emanating from p would intersect the
circumference of each sphere. Now, suppose that on each S,, an area is marked
equivalent to the surface area SO.Then the proportion of lines emanating from
p that will intersect this marked area on sphere SP, would approximate the
k
ratio So& If So is a constant (K), the fraction tends towards ~ as the
4mm2
number of rays approaches infinity. Because the only variable in the expres-
sion is r, then, according to the experiment, the proportion of lines intersecting
an area will decrease inversely as the square of the radius. If we were to follow
the same procedure but reduce the dimensions of the space to two, where we
determine the number of lines intersecting an interval of fixed amount on the
perimeter of various concentric circles, the function describing the decrease
would be l / r . If, however, we were to consider a one-dimensional space, or a
single line as the only path in a two- or three-dimensional space, there would
be no diminution, for there would be no reason to stop the line, unless of course
we assume a friction. But this would not be an assumption of geometry, but
24 / Geographkd Analysis

rather of physical science and would lead us back to our discussion of friction
of distance.
According to this experiment, from the properties of an “empty” Euclidian
space we could expect a “decay” function for rays of lines due entirely to the
dimensionality of space. If instead of rays there were gravitational and electro-
magnetic fields, then aspects of the universe tend to conform to the l/r2 function
in Newton’s law of gravitation and Coulomb’s law of electrostatics. These laws
however are not dependent for their derivations on our mental experiment.l
Furthermore, if the medium separating the bodies has a variable density or an
electrical charge, then the effects of distance in the laws would change. For
geography especially, physical space is not empty and the surface of the earth
is not isotropic. To declare that the use of the inverse square function in physics
justifies its use as a foundation for geography is untenable, for it amounts to
the position that gravitational or electrical concepts are at the basis of geog-
raphy’s explanations. Such is clearly not the case, and to contend that it should
be, is identical to the assertions about the usefulness for geography of the
concept of friction in physics.
There seem to be then no plausible justifications for the claim that properties
of physical space, expressed independently of substances, are useful in explaining
geographic questions. Yet belief in the contrary has tended to make credible,
assertions that geography is a spatial science in search of spatial laws. The
geographer’s dilemma over space is writ large when we compare Bunge’s earlier
statement in Theoretical Geography that geography’s central issue seems to be
“the problem of placing interacting objects as near to each other as possible or
simply the nearness problem” [7, p. 331 to his later modification that “the
central problem of geography is to place interacting objects as near to each
other as possible when the definition of distance is chosen which minimizes total
movement” [7, p. 2001. What then, is the proper definition of distance? Can
we know it without defining the substances of which it is a property?
Certainly the criteria of concept formation in science require that instances
of a concept be identifiable and the concept’s significance be confirmable before
the concept can be judged useful. The former criterion is prior in the sense that
to know if a concept is useful we must first know its instances. However, the
Iatter criterion is the one by which we select from the many possible definitions
of things, those few which become terms of a science. A discussion of the
interrelationshipsbetween these two facets of a concept, and their connection to
explanation may permit us to state more fully the meaning of a relational
concept of physical space and its significance for geography.

111. IDENTIFICATION AND THE BASISOF THE RELA!l’IONAL. CONCEPT

It is important for our purposes to draw attention to two kinds of words in

the natural language system [Z J. The first kind is c d e d a descriptive or refer-

7For a derivation of the inverse square function in Newton’s law of gravitation, see
Bergmann C.3, pp. 216-171.
 Robert David Sack / 25

ential term, such as “John” in the sentence “John is a man.” The second kind
is called a property or character term, as in “man” in the above sentence. The
distinction I wish to draw between these two kinds of words is the difference
in the ways they refer to the occurrence or existence of things. We do not say
that “man” occurs in the same sense of the term as we say that “John” occurs.
One can point and say “That is John,” but one can only say “There is an
instance of a man,” (pointing at John). This distinction is found in the
specialized vocabulary of science where character words are called concepts and
descriptive or referential words refer to instances of concepts or events. “TO
state the Occurrence of events [or to state facts] is to state that a concept has
an instance or a number of instances. ‘Eisenhower is bald’ says that the concept
‘baldness’ has an instance and that this instance is the man we call Eisenhower”
[5,p. 551. But, how can we be sure that you and I are discussing the same
events? How do we distinguish among events?
A process for the unequivocal identification of particulars is essential if
human discourse is to occur. For people who are “near” each other, the une-
quivocal identification and individuation of events and all particulars of human
discourse is ultimately done ostensively: by pointing. The term “ultimately” is
used, for this form of identification is relied on when there is to be absolutely
no doubt about the identity of events. We say “Here, this man is John, not
that one over there”; or “I mean this John, not that John over there.” When
the things to be identified are character terms or concepts, their unequivocal
identification is accomplished by referring or associating them to property terms
and then identifying the latter, and thus indirectly the former, ostensively;
as in “John’s idea.” “What idea?” “The idea that John published four years ago
in Vol. 7, no. 3 of the Journal of Gastronomy, which is found in that public
library.” Ostensive definitions assume that the discussants share the same ex-
periences and sense percepts at the moment the definition is taking place. That,
in a manner of speaking, they are in the same place at the same time. Most
often this is not the case. People are separated and experience different sense
data. Under these conditions, how can the particulars of human discourse be
unequivocally identified? We argue that for identification and individuation to
occur when people have different data or are “separated,” we employ a single
system of physical space and time coordinates and assume: (1) that the system
is knowable to all discussants, (2) that for each and all events we can associate
a unique set of space and time coordinates in the system (from the viewpoint
of absolute space, all events occur ‘‘in space and time”), (3 ) that a concept is
identified by referring it or its instances to instances of other concepts, (4) that
no two events can have the same set of space and time coordinates (or as is
commonly said, no two events can occupy the same place at the same time).8
Within this system, an ostensive definition is not an alternative approach, but
rather a special procedure for associating particular coordinates with particular
events. The single physical space-time system makes it possible to identify and
individuate events that are separate from any single observer and to communi-

8For a discussion of space and time and their relationship to the identification of the
particulars of human discourse see Strawson 1231 and Russell C191.
26 / Geographical Analysis

cate, discuss, or make public, observations that are not immediately apparent
to the senses by everyone. Clearly, these functions cannot be provided by a
plurality of physical spaces, for that would lead to confusion of identification
and perhaps ultimately to solipsism. Only one system of physical space and
time can provide the function of identification and individuation of events.
The use of a single spatial-temporal system to identify the particulars of
human discourse is fundamental. Not only does it appear to be the way
identification is done, but it also appears that such a system is necessary for
identification to occur, in the sense that it is a characteristic of epistemology.
If we agree that events are identified by their association with a single set of
physical space and time coordinates then we arrive a t the contingent fact that
all events have physical geometric properties, if only at a point and for an
instance. The description of these properties would depend on the geometric
characteristics we attribute to the physical spatial system. There is excellent
reason to argue that they should be the properties of a three-dimensional space
of Euclidian geometry with meters or miles as its units.s Only when we are
considering phenomena of sub-atomic or universal proportions would we have
to change our assumption about the appropriateness of Euclidian space. For
the field of geography, these contingencies seem remote.
To speak then of the occurrence of an event is to speak of its physical
geometric properties. We know of the event because with it are associated
physical space and time coordinates. The connection between statements of
events and statements of physical geometric properties of events can perhaps
be made more precise if, instead of juxtaposing the term “event” with physical
geometric properties of events, we refer to a division in the vocabulary of science.
On the one hand there are terms in science that are geomtric, and on the other
we have all other terms (apart from logical ones) which we shall refer to as
substance terms. The word “event” will be neutral, not belonging to either set
of terms, yet signifying the combination of both substance and geometric terms.
To state that an event occurs is to imply that instances of both substance and
geometric concepts occur.
More precisely, then, we can say that for every instance of an event defined
in the substance part of our language there is associated with it an instance of a
physical geometric concept. The critical question about space is how do we
determine if instances of spatial concepts occur. This query places us in the
midst of the philosophical controversy about absolute space.
The use of physical spatial terms in geography independent of substance
referents may have as its foundation the conception of a three-dimensional
Euclidian space of the physical world as a highly structured uniform frame of
reference, which is contingently related to the physical world by having at least
two of its coordinates (instances of geometric terms) referred to proper names or
places (instances of substance terms [19,pp. 242-491). The l o g i d properties
of the geometry are such that after the origin is agreed upon and two coordinates
9 If we distinguish between a “pychoIogical” space or a space of our senses and the space
of physics, the latter, not the former, satisfies the criteria necessary for public identification
and individuation of events.
 Robert David Sack / 27

are named, all other points in the space are assigned unique coordinates which
are ordered by the rules of the Euclidian system, thus an instance of physical
space would be individuated and have a meaningl by being referred to specific
coordinates. This meaning, of physical space (which can be expressed as part
of the classical concept of absolute space) will be termed an unmodified meaningl
of physical space. It implies that instances of physical geometric terms need
refer only to instances of physical space: that there need be no connection
between terms of the space and substance terms. Physical space can occur
independently of, and apart from, substances. It is everywhere the same, immov-
able and absolute. Apart from the arbitrary assignment of coordinates, there
is no way one part of the space could be differentiated from another. Such could
be the meaning of physical space suggested by the use of an unmodified
geometric term in a geographic hypothesis or sentence.
The unmodified meaning of physical space however resurrects the well-known
controversy over the adequacy of absolute space as a concept in science, for it
seems to leave unanswered the question about how we are to observe instances
of absolute space. What does it mean to say that physical space occurs inde-
pendently of substance? Does the term (‘occur” apply in the same way to
physical spatial instances as it does to instances of substance terms? These
queries are especially urgent if we assume that to unequivocally identify the
occurrences of instances of substance terms requires that they be uniquely
associated with instances of physical spatial terms. Therefore, there are a t least
some physical spatial terms that are associated with substance terms. If, how-
ever, space is absolute and the unmodified meaning of physical space is
sufficient, then there would be an asymmetry in the requirements for observa-
tion and identification of spatial and substance terms. The term ‘(occur” in both
cases would have different connotations.
Concern with these issues in physics helped occasion the introduction of the
concept of an ether as a way of substantiating absolute space. The ether was
thought to be composed of uniform particles distributed in space and thus an
instance of space would refer to an instance of these particles. However, the
ether raised the same problems as were encountered with absolute space. We
could not differentiate one part of the ether from another and would have no
way of determining an instance of it.
With the advent of relativity, the view of absolute-empty space was severely
challenged. Interpretations of relativity have led to the claim that physical
space is not empty and that the structure of physical space is determined by
the concentration of matter and energy. Space and substance (matter and
energy) can be considered interconnected and the question about the occurrence
of space, although by no means resolved, is a t least thought to be dependent on
the questions about the occurrence of substances.
It is our contention that the rudiments of this interrelationship between space
and substance can be expressed on a semantic level as the reciprocal conjunction
between instances of physical geometric terms and instances of substance terms.
Specifically, we claim that there is a more extended expression of meaning, of
physical space-a modified meaning, of physical space-which is based on the
28 / Geographical Analysis

supposition that for every instance of a physical spatial term there is correspond-
ing to it, or associated with it, an instance of one or more substance terms. (Or,
in the terminology of the unmodified meaning of space, for every set of coor-
dinates in space there corresponds one or more instances of substance terms.)
According to this modified meaning, of physical space we cannot make a state-
ment about physical space without making or implying statements about
substance terms. If, for example, we are in an office and point to a length of
ten feet, we could be referring to the geometric properties of such things as ten
feet of floor, or wall, or ceiling. If we were to point to an area in an ,‘empty’’
room, the area could refer to an amount of air or a volume of gas molecules
called air. If an astronomer were to refer to two points, located between two
stars in outer space, and the “line” connecting them, the line would represent
the length of part of a light ray or a gravitational field. The modified meaning
of physical space then implies that space is not empty and that there is no
action a t a distance.
The modified meaning of physical space allows for a completely reciprocal
connection between substance and physical geometric terms, and thus a sym-
metry in the way we assert their occurrences. We say that an instance of space
occurs when it can be connected or related to an instance of substance terms
and we say that an instance of a substance term occurs when we can connect it
with instances of space (and time). We would not use the term occur for
instances of substances that cannot be located in physical space, or for instances
of physical spatial concepts that have no substance referents. Such a reciprocal
relationship agrees also with the definition of a synthetic geometry as a system
for which we have a text so that any conceivable instance of a non-logical
geometric term can be referred to terms in another language system, namely to
the substance terms in science. Furthermore, the modified meaning of physical
space is borne out in our own language system where we speak of something
having a length or something having a shape. The somethings refer to sub-
stance terms.
We argue then that there are two ways in which we can express the meaning
of physical space, or assert that instances of physical space occur. The first,
which is an unmodified meaning, requires only that the Euclidian coordinate
system be connected to at least two terms in the substance vocabulary and that
an instance of a physical spatial concept be identified as a set of coordinates
in this system. The modified meaning of physical space allows that individuation
can occur in this way, and adds that to each and every instance of a physical
geometric term there are associated with it corresponding instances of substance
terms. The adequacy of these two meanings of physical space depends on the
sentences in which the physical geometric terms are to be employed. The
modified meaning of physical space forms the basis of the relational concept,
which asserts that for all but the laws of physical geometry, the modified
meaning of physical space must be used and the significant substance referents
of the geometric terms must be made clear in the hypotheses. The justification
of the relational position rests on the requirements that must be met for a
geometric term to have significance in a non-geometric law.
 Robert David Sack / 29

IV. EXPLANATION
AND THE RELATIONAL
CONCEPT

Paralleling the distinction between geometric and substance terms are two
kinds of laws. The first are laws of physical geometry which (apart from logical
terms) need contain only physical geometric terms. Its laws are about the
physical geometric structure of all events and except for the sub-atomic or the
astronomic, their validity does not depend on the substances examined. But the
laws of physical geometry are static, and furthermore, they cannot be deduced
from or explained by laws of process [ 2 0 ] . They form a relatively closed and
complete deductive axiomatic system.
The second kind of law, which will be called a substance law, includes, with
trivial exception, all of the remaining laws of science, ones that are both static
and dynamic and ones that may or may not include geometric concepts.l0
Substance laws must contain explicit mention of substance terms and explain
and predict their instances. Unlike the laws of physical geometry, we can say
that any static substance law can in principle be explained by or deduced from
dynamic laws, if such laws are known.
Because the identification of instances of substance terms requires ultimately
that they be located in space and time ( a requirement which would be met
whether we assume a modified or unmodified meaning of physical space), a
physical geometric property of an event can be explained in two different ways,
either by physical geometric laws or by substance laws [ 2 0 ] . Suppose we want
to explain the geometric property of an event such as a line d18 feet long which
is the length of the hypotenuse of the top of my desk. To explain it we can state
that the line is an hypotenuse of a right-angle triangle whose sides are both
three feet long and that, according to a familiar geometric law, the square of the
hypotenuse of a right-angle triangle is equal to the sum of the squares of the
other two sides. Or we can explain the occurrence of the event by considering
it to be an instance of a substance concept in a substance law explaining the
occurrence of desk tops, i.e., whenever there is a white room with a skylight,
there is a desk of this size and shape in the room. The first explanation considers
the event to be an instance of a concept of physical geometry. The second
considers it an instance of a substance concept in substance laws, which may
include physical geometric concepts but which are not laws of the geometry of
physical space. If the substance laws are static, then there is no reason why in
principle they cannot be related to dynamic laws, if such laws were known.
Although physical geometric laws explain the occurrence of physical geometric
properties of events, they are in principle static and cannot be connected to laws
of process.
Geographic questions are in part about the physical geometric properties of
events. To the extent that geographers want to explain the causes of these

1 0 1 wish to exclude “cross-connection” laws and such laws as “green is extended” from
the definition of a substance law. For a discussion of explanation in geography, consideration
of these two forms of laws is unnecessary. If we define substance laws as all remaining laws
of science that are not geometric laws, then all substance laws may in principle be derived
from or explained by dynamic laws C201.
30 / Geographical Analysis

geometric properties of events and connect them with other dynamic laws of
science, physical geometric laws alone cannot satisfactorily answer geographic
questions. They answer different questions about the occurrence of physical
geometric properties of events, ones that are locked within a relatively complete
and consistent deductive axiomatic system. All substance laws, whether they
are static or dynamic, or whether they contain physical geometric terms or not,
explain the occurrence of physical geometric properties of events by virtue of
the contingent relationship between substance concepts and properties of phys-
ical space. As long as geographers do not preclude specific substances from their
purview, all substance laws can explain aspects of geographic questions.
Although there is no logical necessity to include physical geometric terms in
substance laws, there are numerous substance laws in geography and other
disciplines that contain them. But substance laws are about substance concepts.
Unlike the terms of physical geometry, substance concepts are not delinitionally
or logically interconnected by a simple rule such as a metric. Therefore, the
interpretation of the meaning of physical geometric terms in substance laws is
critical. It is the position of the relational concept of space that a necessary
condition for a physical geometric term to be judged useful (have a meaningz)
in a substance law is that the modified meaning, of physical space be employed
where the substance referents of the geometric terms are named. Unmodified
physical geometric terms are incomplete concepts in substance laws. Each law
must make clear to what substances its geometric properties refer.
If we examine the physical sciences and look a t the non-geometric laws which
explicitly include geometric terms, we notice that in the application of the laws
we are able to determine to what substances the geometric terms refer and that
changes in the substances would make a difference in the statements of the laws.
We know for instance that there are no frictions of distances but rather coeffi-
cients of friction, each depending on the substances involved; that the force of
attraction between two electrically charged bodies depends on the distance and
the nature of the substance separating the bodies; that the speed a t which sound
will travel between two places is dependent on the length of substances separat-
ing these places. There are of course many occasions when the substance
referents of a physical geometric term are clear from the context of the problem
and need not be mentioned explicitly as a condition of the law. In the law
explaining the period of the pendulum for instance, distance is introduced as a
variable in the form of the length of the material that suspends the bob.
Although the law makes no explicit mention of the substance of which this
is a length, knowning the material is essential for implementing the law, both
the mass and the thickness of the material will affect the pendulum’s periodicity.
In Newton’s law of gravitation (which is the model for Stewart’s and Warntz’s
statements about social physics), the force between two bodies is directly pro-
portional to their masses and inversely proportional to the square of the distance
between them. Here also, it may appear as though distance were considered
independently of events. But, on the contrary, the nature of the medium or the
substance which separates the bodies could make a significant difference. If two
planets were separated by a fluid of uneven density, the determination of the
force between them could not be solved precisely by Newton’s law. Often the
 Robert David Sack / 31

law of gravity is said to apply to conditions in vacuum, or under conditions

where substances are absent. Whether or not there can be emptiness, is a critical
issue for the philosophy of space. Yet, it seems that thus far scientists have not
been able to create a perfect void, and according to the theory of relativity and
current conceptions of substance, the universe is everywhere full of either matter
or energy. The fact that two bodies could attract each other means that there
are fields of energy between them, not empty space. The reason that substance
terms need not be explicitly connected to geometric terms in these laws is that
we know more than these laws tell us; we have other laws that determine what
the effect would be if the substances or the “context” were changed.
I n the social sciences there are far fewer examples of laws. But to illustrate
the point, suppose after intensive study of a well-defined, statistically significant
sample of people, placed, two at a time, in a quiet, clean, well-lighted, empty,
ten-foot by ten-foot room, where visibility was excellent, it was found that when
these pairs of people carried on a conversation they stood, on the average, three
feet apart. The results of the experiments may be expressed in a generalization
or “law” that under these conditions, three feet is the normal conversational
distance. Here the expression uses the unmodified distance term three feet. But
clearly it is three feet of clear air that is separating these people. The only
reason that the generalization makes sense using the distance expression unmod-
ified, is that we know more than the generalization tells us. If, for example,
the speakers were standing three feet apart and there was a one-foot-thick pane
of glass between them, a conversation would not be likely, nor would we expect
it to be, because we know that sound does not travel as well through glass as it
does through air. If there were no pane of glass between the speakers but
instead the air in the room was heavily foggy, then the distances separating
the speakers may be less than if the air were clear. But all of this would not be
startling. Nor would it compel us to state explicitly in the “law” that it is a
distance of three feet of clear air that separates the speakers and not panes of
glass or foggy air. The reason we can omit the referents to clear air (or exclude
the referents to panes of glass) is that we know enough about the laws of
physiology, the physics of sound and light, and we have had a sufficient
familiarity with conversations ourselves, to realize how significant are the sub-
stance referents “clear air” or “panes of glass.” I n most of the inquiries in
geography, especially human geography, however, the context is far from under-
stood. Nonetheless, physical geometric terms are often used, unmodified, without
substance referents. We look for distance decay functions, for the effects of
friction of distances, for action spaces, for awareness spaces, and for marriage
and migration fields. The significant substances may not be known and often
only vague references are made to the context in which the behavior took place.
Yet distances or areas are presented as variables in themselves, with the expecta-
tion that these geometric concepts and others may become significant in sub-
stance laws.”

11Geography’s use of unmodified physical spatial terms may stem from a belief in the
groundless assertion that a law must contain them for it to be “geographic” or “spatial” or
“an explanation of a geographic question.” While appearing to give geography some dis-
c i p l i n q independence, such an assertion when put into practice, leads to the use of incom-
32 / Geographical Analysis

Several words of caution about the implications of the relational concept of

physical space are in order. The relational connection between geometric prop-
erties and phenomena may suggest that all substance terms can be replaced by
terms of physical geometry, and then we would need only one set of laws to
explain these events, namely the geometry of the physical world. This conclusion
is not tenable. The translation of the terms of the substance language into
physical geometric terms would never be complete. There must remain at least
two terms outside of the geometric language to identify the origins of the spatial
system [ 29, pp. 242491. Even if the translation were attempted, it would be,
to say the least, extremely laborious, with perhaps the only gain being some
facility in the process of individuation [ 17,261. Most importantly, it would not
abolish the distinction between geometric and non-geometric laws. The only
difference would be a change in the terms (semantics), not the structure (syntax)
of the substance laws. Events would still be perceived by their substances, such
as colors, smells, or tastes, and we would still seek laws about the occurrences
of these events. A geometric language would only change the naming of terms
or concepts, and perhaps their individuation, but not their explanation.
Furthermore, a rejection of absolute physical space and the adoption of a
relational one, does not mean a rejection of Newtonian physics. The functioning
laws of classical physics do not require the assumptions of absolute space.
Absolute space was necessary only if Newton’s laws were to be the laws of the
entire universe. Within the framework of inertial systems, Newtonian physics
becomes appropriate for a particular system but not for all. I n this respect,
Newton’s laws are not universal or absolute but are relative to particular frames
of reference that happen to be of interest to us, and no conception of absolute
space needs to be included [II, p. 139; 24, p. 116].12

CONCLUSION

The commitment we have to a particular concept of physical space in our

discipline has extreme import, practically as well as philosophically. The
adoption of a relational concept of space in geography may mean a significant
change in current conceptions of geography and geographic explanation. From
the relational concept of space and the structure of scientific explanation follow
three positions concerning geography and explanation: (1) all substance laws,
whether or not they explicitly include physical geometric terms, could explain
aspects of geographic questions, for geographic questions are concerned with
why events or physical geometric properties of events occur. (2) If physical
geometric terms are included in substance laws they must be explicitly con-
plete spatial concepts, and to the quandary over the proper measure of geographic space, all
of which make the empirical testing of many geographic propositions an intractable problem.
Furthermore, acceptance of the assertion may restrict the search for explanations to geo-
graphic questions and may result in sterile arguments about the nature of geography.
12It should also be borne in mind that our rejection of absolute space was made from
the viewpoint of its usefulness as a concept in science, espedally a science of geography.
However, from the viewpoint of metaphysics and even, it seems, from some philosophical
interpretations of theoretical physics, implications of absolute space are still not clear 191.
 Robert David Sack 33

nected to non-geometric (substance) terms. The law (or its context) must tell
us to what phenomena the physical geometric properties are connected. Thus
in geographic “explanations” there can be no significant ((spatial” variables
without named substance referents. (3) The naming of the substance referents
of a geometric term does not necessarily mean however that this geometric
property of an event will be a significant concept in an hypothesis. It is a matter
of fact whether a geometric property of an event is significant in a non-geometric
law.
Together these statements about geography and geographic explanation sug-
gest that the apparent generality and deductive rigour of many of the “spatial”
formulations in our field have been due to the omission of the referents of the
spatial concepts and to the assertion that physical spatial properties are a
priori useful. This omission has paid its toll in the clarity and accuracy of our
statements, a price that is itself the value of science. Furthermore, to justify
these formulations, geographers have referred to their field as a “spatial”
discipline, based on “spatial” laws, implying that geography is a “spatial
science” apart from other sciences. But a relational concept of physical space
raises serious objections to these ~ 1 aim s .l~
According to the relational concept,
any substance law (whether static or dynamic) could be a candidate for an
explanation of a geographic question. To ask if things occur is to ask if they
occur somewhere, and to explain why they occur, is to explain why they occur
where they do.
The relational concept may help clarify statements in other sciences which
assume that properties of physical space are useful in explaining human be-
havior, but which have not stated clearly what is meant by physical space. For
instance the arguments about overcrowding or territoriality are often presented
without specifying the referents of the spatial terms. From them we have dif-
ficulty knowing if overcrowding is meant to refer to density, or to privacy, or
if animals are supposed to defend areas of space independently of events, or
if the amount of territory they claim will depend on the events within this area.
The meaning of these and other formulations depends in large measure on the
meaning of physical space. The only way for properties of physical space to be
introduced into non-geometric hypotheses and still comply with the rules of
concept formation is to name in the hypotheses to what substance terms the
spatial terms refer. Only then are non-geometric laws including a meaning of
physical space that complies with the tenets of science.

LITERATURE CITED

1. h u s , J., R.
ABLER, and P. GOULD.Spatid Organization: The Geographer‘s View of
the World. Englewood Cliffs, N.J.:Prentice-Hall, Inc., 1971.
2. BERGMANN, G . Philosophy of science. Madison, Wisc.: University of Wisconsin Press,
195’7.
3. Bmcmm, P. G. The Riddle of Gravitation. New York: Charles Scribner’s Sons, 1968.

13Serious objections to these claims have becn raised from other viewpoints by Blaut
C41 and May 1131. See footnote 3.
34 / Geographical Analysis

J. ‘‘Space and Process,” The Professional Geographer, 13 (1961), 1-7.

4. BLAUT,
5. BRODBECK, M. “Logic and Scientific Method in Research on Teaching,” in N. L. Gage
(ed.) Handbook of Research on Teaching. New York: Rand McNally, 1963.
6. -. “Meaning and Action,” Philosophy of Science, 33 (1963), 309-24.
7. BUNCE,W. Theoretical Geography. Lund: C.W.K. Gleerup, 1966.
8. CARNAP, R. “Formal and Factual Science,” in H. Feigl and M. Brodbedc (eds.) Read-
ings in the Philosophy of Science. New York: Appleton-Century-Crofts, Inc., 1953.
9. GRUNBAUM, A. “The Philosophical Retention of Absolute Space in Einstein’s General
Theory of Relativity,” The Philosophical Review, 66 (1957), 525-34.
10. HARVEY, D. Explamtion in Geography. London: Edward Arnold, Ltd., 1969.
11. JXMMER, M. Concepts of Space. New York: Harper and Brothers, 1960.
12. LUKERMA“, F. and P. PORTER. “Gravity and Potential Models in Economic Geography,”
A n d , Association of American Geographers, 50 (1960), 493-504.
13. MAY, J. A. Kant’s Concept of Geography and Its Relation t o Recent Geographical
Thought. Toronto: University of Toronto Press, 1970.
14. M o m , R. L. and F. R. PITTS. “Marriage, Migration and the Mean Information Field:
A Study in Uniqueness and Generality,” Ann&, Association of American Geographers,
57 (1967), 401-22.
15. M o m , R. L. T h e S M t i d Organization of Society. California: Wadsworth, 1970.
16. OLSSON, G. Distance and Human Interaction. Philadelphia: Regional Science Research
Institute. 1965.
17. QUINE, W. V. “Identity, Ostention, and Hypostasis,” The Journal of Philosophy, 47
(1950), 621-33.
18. RUSHTON,G. “Analysis of Spatial Behavior by Revealed Space Preference,” Ann&,
Association of American Geographers, 59 (1969), 391-400.
19. RUSSELL,B. Human Knowledge: Its Scope and Limits. New York: Simon and Schuster,
1948.
20. SACK,R. D. “Geography, Geometry, and Explanation,” A n d , Association of American
Geographers, 62 (1972), 61-78.
21. STEWART, J. Q. “Population Projection by Means of Income Potential Models,” Papers
and Proceedings, Regional Science Association, 4 (1958), 153-54.
22. STOUFFER,S. A. “Intervening Opportunities: A Theory Relating Mobility and Distance,”
American Sociological Review, 5 (1940), 8 4 5 6 7 .
23. STRAWSON, P. F. Individuals: A n Essay in Descriptive Metaphysics. New York: Double-
day, 1963.
24. VAN FRAASSEN, B. C. An Introduction t o the Philosophy of Time and Space. New York:
Random House, 1970.
25. WARNTZ,W. “Geography of Prices and Spatial Interaction,” Papers and Proceedings,
Regional Science Association, 3 (1957), 118-29.
26. WILSON,N. L. “Space, Time, and Individuals,” The Journal of Philosophy, 52 (1955),
589-98.
27. YEATES,M. H. and B. J. GARNER. The North American City. New York: Harper and
Row, 1971.