The Fibonacci Quarterly: Number

r^
THE FIBONACCI QUARTERLY

df^^^fxÂij^
"""^Cl^
THE OFFICIAL JOURNAL

THE FIBONACCI ASSOCIATION
VOLUME 8
NUMBER
CONTENTS
PART IADVANCED
A Fibonacci Circulant.
Combinatorial Problems
for Generalized Fibonacci Numbers .
Application of Recursive Sequences
to Diophantine Equations .
D, A . Lind 449
Verner E. Hoggatt, jr. 456
Raphael Finkelstein, Edgar Karst, and Hymie London 463
On a Class of Difference Equations

On a Conjecture of Dmitri Thoro . .
The Fibonacci Numbers
Considered as a Pisot Sequence .
Modulo One Uniform Distribution of
of Logarithms of Certain Recursive
Advanced Problems and Solutions.
RaymondE. Whitney 470

David G. Beverage 475
.
Morris Jack DeLeon 476
the Sequence
/. L. Brown, jr.,
Sequences . . . . and R. L. Duncan 482
*
Edited by Raymond E. Whitney 487
L e t t e r to the E d i t o r
David G. Beverage 498

PART II -
ELEMENTARY
Determination of Heronian Triangles
John H. Carlson 499
Pineapples and Fibonacci Numbers . . . . . . .
Philip B. Onderdonk 507
The Powers of Three
J.M. Williams, Jr. 509
Some Remarks on the Ordering

of General Fibonacci Sequences . . . . . . . . .
The Educational Value in Mathematics
Certain Arithmetical Properties of ^k(ak + 1)
Joe R. Crenshaw 516

John B. Lewis 522
. Joseph Arkin 531
Fibonacci Numbers as Paths

of Rooks on a Chessboard .
Edward T. Frankel 538
Elementary Problems and Solutions
Edited by A . P. Hillman 542
Volume Index
552
DECEMBER
1970
THE FIBONACCI QUARTERLY

THE OFFICIAL JOURNAL OF THE FIBONACCI ASSOCIATION
DEVOTED TO THE STUDY
OF INTEGERS WITH SPECIAL PROPERTIES
EDITORIAL BOARD
H. L. Alder
Marjorie Bicknell
John L. Brown, J r .
Brother A. Brousseau
L. Carlitz
H. W. Eves
H. W. Gould
A. P. Hillman
V. E. Hoggatt, J r .
Donald E. Knuth
George Ledin, J r .
D. A. Lind
C. T. Long
M. N. S. Swamy
D. E. Thoro
WITH THE COOPERATION OF
T e r r y Brennan
Maxey Brooke
Paul F. Byrd
Calvin D. Crabill
John H. Halton
A. F. Horadam
Dov Jarden
Stephen Jerbic
L. H. Lange
James Maxwell
Sister M.DeSales McNabb
D. W. Robinson
Azriel Rosenfeld
Lloyd Walker
Charles H. Wall
The California Mathematics Council
All subscription correspondence should be addressed to Brother Alfred

Brousseau, St. Mary T s College, California. All checks ($6.00 per year)
should be made out to the Fibonacci Association or the Fibonacci Quarterly.
Manuscripts intended for publication in the Quarterly should be sent to
Verner E. Hoggatt, J r . , Mathematics Department, San Jose State College,
San J o s e , California. All manuscripts should be typed, double-spaced.
Drawings should be made the same size as they will appear in the Quarterly,
and should be done in India ink on either vellum or bond paper. Authors
should keep a copy of the manuscript sent to the editors.
The Quarterly is entered as third-class mail at the St. Mary T s College Post
Office, California, as an official publication of the Fibonacci Association.
A FIBONACCI CIRCULANT
D. A. LIND
University of Virginia, Charlottesville, Virginia
1. Put
1 F
r
F
F
D
n,r
r+n-1
r+n-2
r+2
'r+n-1
r+1
"r+n-2
F
r+n-1
' Fr+1
where F
r+1
F
r
r+2
r+n-3
r+3
denotes the Fibonacci numbers defined by

J
n
Yi = Fo = 1.
1
2
*
= F
n+2
+ F
n+1
We show that
(F
(1)
D
n,r
(F
- F
)
1
;
"n+r
n+r-1
r
l
1
1 - L + (-1) "
n
where L
= F - + F , - is the n
Lucas number*
n
n-1
n+1
A circulant is a determinant of the form
a
n-1
(2)
C(a0,"- ' a n - l } =
n-2
l
0
Vi
a
a
2
l
S O S
Vi
see
s a
an-3
0
n-2
e
e
i
449
9 9
o|
450
Dec.
It is known (see [1, Vol. 3 9 pp. 374-375] and [3, p . 39]) that
n-l/'n-l
. \
(3)
where the
,,
2k?r , .
2k7T
Xt) = cos
+ i sin
k
n
n
th
are the n roots of unity* To establish (3) rapidly s multiply
Ctao,.-..^)
bytheVandermonde determinant V = Jco?I (i s j = 0S 1, 8 8 , n - 1). Denoting

the right side of (3) by P t by factoring out common factors, one finds CV =
PV, and since V ^ 0, (3) follows.
Now D
is a special case of (2) with
n9 r
a. = F . ,
j
j+r
J+r
in which a = (1 + N / 5 ) / 2 , /3 = (1 -y/E)/2.
- fl>+r
a - 0
Thus by (3),
Ji+r +
, r[ar - 1 - ,jc-l
ji+v-1-.Jo,
n-1- a r - /3r - an+r +, (r
/3
-a n+r-1 ,+/S
=
k^0
n X
" k =o
" " > - 0 ( 1 - 0 k ) ( l - /3w k )

F
r " F n + r " < F n+r-l " F r - l ) W k

- T i - <*>kHi - ^ )
1970
451
Now for any x and y ,
(4)
Ivx-y^-yS(!-"*)-4"
= x
n
- y
Therefore
CFr " F n + r " ( F n + r - l ~
r-l)C\l
(F
'
r " W
"
" <Fn+r-l " V l * "
and
n-1
n-1
n-1
n a - aoj ){i - pw ) n a - aoj ) n (i = (1 - an)(l

w h e r e we have u s e d L M = a + p
- ^)
JSW.
1 - L n + (-l)n
simplifies if n i s even* Ruggles
n i ic
[ 2] h a s shown that
F _,_ - F
n+p
It follows that if n = 0 (mod 4 ) ,
L F ,
n p
p even
F L ,
n p
odd
,nfTn
-i
2 - L
and if n s 2 (mod 4) s
n r n
D
n9r
T h i s e s t a b l i s h e s (1).
We note that
that t h i s evaluation of D
>
vn
r 1+ in
-i
H M 2 - ~L - 2 J
452
[Dec.
2. The generalization of (1) to second-order recurring sequences uses

the same techniques. Consider the sequence { w } defined by
n+2
* n+1
^ n
WQ and Wj arbitrary, where p 2 - 4q ^ 0. Let a and b be the roots of

the auxiliary polynomial, so that a ^ b and ab = q. We shall assume that
neither a nor b is an n
root of unity* Since the roots are distinct, there
are constants A and B such that W

{V n } by
= A a + Bb .
Define the sequence
Vn = a + b .
Put
D
(W) = C(W , W ^ , ' ' , W ^ J
n,r
r
r+1
n+n-1
Setting a. = W. + r = Aa : i + r + Bb^ +r in (3) gives
Dn
n , r
n-1 / n - 1
.
A
(W) = n ( Aa r (ao). ) j + Bbr(bco ) ] )
K
K
k=0 \ j=0
/
- V /*a r (l - a">
k=0 \
n
n
k=0
r -
aw
Bbr(l - b n ) \
1
bw
n+r " ^ W r - 1 ~ W n + r - l ^ k
(1 - a c k ) d - b a y
(W r - W n + r ) n - q n (W r -l - W ^ ^ ) 1 1
1 - Vn
qn
which agrees with (1) by taking p = 1, q = - 1 , W = F , and V
3, We now consider a slight variant of the above.
Put
= L .
1970]
453
r+1
F
r+n-1
E
r+n-2
nfr
-F
r+n-1
r+1
r+n-2
r+n-3
-F
r+2
r+1
r+n-1
"r+2
F
r+3
We shall prove
(Er + F n+r
^_ f
(5)
n,r
+ (-l) n (F n+r-1
^ i + F r - 1 .)n
1 + L + (-l) n
n
A determinant of the form
V"'Vi
*n-l
is termed a skew circulanfc,
"an-2
"Vl
"al
~a2
n-2
n-3
Scott [1, Vol. 4, p. 356] has shown that
n-1 / n-1
. \
where the
cos
n-l
~a3
(6)
k =
(2k + l)ir __, . . (2k + l)ir

1 J - + i sin 2
454
[Dec.
a r e the n
th
r o o t s of - 1 . To p r o v e (6) quickly, multiply S(a 0 , , a
the Vandermonde d e t e r m i n a n t | e] | (i, j = 0S 1, * , n - 1),
) by
and t r e a t a s in
the proof of (3).

To evaluate E
n,r
l e t a. * F . , .
j
j+r
A development s i m i l a r to Section 1
F
shows that
n - 1 / n - 1 aT(a, ) j - /3r(j3, );
a - j8
k=0 \ j =0
(7)
n-1
n+r
+ F
+ TF , - + F
-If.
r
L n+r-1
r-U k
(1 - ark )(l - iSk )
"
k=0
F o r a r b i t r a r y x and y ,
(8)
Iv^'-SO^M
+ 1
n _, n
+ y
Application of this to (7) yields the d e s i r e d r e s u l t (5).

We r e m a r k that a s before (5) simplifies for even n.
shown that
F ^ + F
n+p
n-p
F L ,
n p
L F ,
n p
p even
^
p odd
F
Then if n s 0 (mod 4),
_ ^(*Hn
nsr
+ F
2 + L
r-l+|n)
Ruggles [ 2 j h a s
1970]
455
while if n = 2 (mod 4),
|nl
r+|n
r-l+|nj
E
2 + L
n
Note that the l a t t e r y i e l d s on c o m p a r i s o n with the d e t e r m i n a n t the identity
5(F 2 , + F 2 ) = L 2
+ L 2 = 5F Q _,_r+1
r
r+1
r
2r+l
4.
The extension of this to s e c o n d - o r d e r r e c u r r i n g s e q u e n c e s involves
no new i d e a s , and the d e t a i l s a r e t h e r e f o r e omitted.
and V be a s
n
th
before j with the exception that we r e q u i r e a and b not be n
r o o t s of - 1
r a t h e r than +1 to avoid division by z e r o . P u t
E
n,r
Let W
(W) = S(W , W ^ , , W ^ , ) .
r
r+1
r+n-1
Using (6) and (8), we find
(W)
(W
n+r
) +
n ( W
n+r-l
1+V
n
which r e d u c e s to (5) when q = - q = 1,
r-l)n
11
H-q
^
= F , and V
= L
REFERENCES
1.
T h o m a s M u i r , The T h e o r y of D e t e r m i n a n t s (4 V o l s . ) , D o v e r , New York,

I960.
2* L D. R u g g l e s , "Some Fibonacci R e s u l t s Using F i b o n a c c i - T y p e S e q u e n c e s / '

Fibonacci Q u a r t e r l y , 1 (1963), No. 2, pp. 75-80,
30
V* I. S m i r n o v , L i n e a r A l g e b r a and Group T h e o r y , M c G r a w - H i l l , New

Y o r k , 1961.
COMBINATORIAL PROBLEMS
FOR GENERALIZED FIBONACCI NUMBERS
VERNER E. HOGGATT, JR.
San Jose State College, San Jose, California
Theorem 1. The number of subsets of ( l , 2, 3, , n} which have

k elements and satisfy the constraint that i and i + j (j = 1, 2, 3, * , a)
do not appear in the same subset i s
fci ( n ik )\
a '
(0
where I
subset.
/ n - ka + a\
=\
] is the binomial coefficient.
We count </), the empty set, as a
Comments. Before proceeding with the proof, we note with Riordan

[ 1 ] , that for a = 1, the result is due to Kaplansky. If, for fixed n, one
sums over all k-part subsets, he gets Fibonacci numbers,
[<n-H)/2]
n k +1
*+1- L ( \
>
tai<
k=0
where [x] is the greatest integer function.
The theorem above is a prob-
lem given in Riordan [2],

Proof. Let g (n,k) be the number of admissible subsets selected from
~a
the set ( l , 2, 3, , n}. Then
y (n + l,k) = g (n,k) + g (n - a, k - 1) ,
d.
<x
<x
since g (n,k) counts all admissible subsets without element n + 1 while

a
g (n - a, k - 1) counts all the admissible subsets which contain element
a
n + 1. If element n + 1 is in any such subset, then the elements n, n - 1,
n - 2 , n - 3 , , n - a + 1 cannot be in the subset.
We select k - 1 ele-
mentsfromthe n - a elements 1, 2, 3, , n - a to make admissible subsets and add n + 1 to each subset. The count is precisely g (n - a, k - 1).
a
456
Dec.
1970
COMBINATORIAL PROBLEMS FOR

GENERALIZED FIBONACCI NUMBERS
Consider
<a<". = ( n - i r a ) .
457
k > 0
But, since the fa (n,k) are binomial coefficients s
n + l - k a + a\
f a (n + l,k)
[ n - k a + a \ / n - a - (k-l)a + a\
j {
k-i
)
k
; -\
f (n,k) + f ( n - a, k - 1)
d
Thus, fa (n,k) and g a (n,k) satisfy the same recurrence relation. Since the
boundary conditions are
5 a (n,l) = f a (n,l) = n ,
and
g (l,n) = g_(l,n) = 0,
d
n > 1 ,
II
the arrays are identical. This concludes the proof of Theorem 1.

We note that 5 for fixed k > 09 the number of k-part subsets of
( l , 2, 3, ., n} for n = 0, 1, 2, , are aligned in the k
1
Pascal s left-adjusted triangle.
column of
If one sums for fixed n the number of k-
part subsets, one obtains
V (n,a)
d
[n-fal
fn+al
a+l'J
La+lJ
I^=A
k=0
lr=A
k=0
where [x] is the greatest integer function.
'
These are precisely the gener-
alized Fibonacci numbers of Harris and Styles [3]. There,
458
[DecB
[n/(p+D]
u(n;p,l) =
k=o
k=0
so that
Va( n , a ) = u(n + a; n , l )
Clearly, if we select only certain k-part subsets (b > 1)
TiH-al
La+bJ
T.h... - Z (" 'I? + S )

k=0
then
Va (n,a,b) = u(n + a; a,b) .
Thusj one has a nice combinatorial problem in restricted subsets whose
solution sequences are the generalized Fibonacci numbers defined in [ 3 ] and
studied in [4], [5], [6], [7], [11], and [12],
GENERALIZATION
We extend Theorem 1 to all generalized Pascal triangles,
Theorem 2. The number of subsets of ( l , 2, 3, , n} with k elements in which i, i + j (j = 1, 2, , a) are not in the same subset nor
are simultaneously all of the integers i + j a + 1
in the same subset, is
( i \ = jj n - ka + a i^
fca(n,k,r)
k
where
(j = 0, 1, 2, , r - 1 ) ,
1970]

n(r-l)
r 1
(1 +x +x + +x - )
459
i n-f , X- 1
=
i=0
We call
n
i,
the r-nomial coefficients, and n designates the row and i designates
the
column in the generalized Pascal triangle induced by the expansion of

(1 + x + x2 + . - . + x r
i
X
) , n = 0,1,2," .
Proofs Let g (n 9 k s r) be the number of admissible subsets selected

with elements from {l, 29 3, , n}. Then
3 ,
g a (n + l,k,r) = ga.(n,k,r) + ga.(n - a,k - l 5 r ) + gd (n - 2a,k - 2,r)

+ + g (n - (r - l')a, k - r + l,r)
a
Consider the set of numbers n + 1, n - a + 1, n - 2 a + l , n - 3 a + 1, * *
n - (r - l)a + 1. The general term g (n - sa 5 k - s 9 r) gives the number of
a
admissible subsets which require the use of n + 1, n - a + l 5 n - 2a + 1
' , n - ( s - l)a + 1 ? disallows the integer n - sa + 1 5 but permits the use
of the integers n = 1, 2, 3 ? , n - sa in the subsets subject to the constraints that integers i, i + j (j = 1, 2, 3, * e , a) do not appear in the
same subset,, This concludes the derivation of the recurrence relation.
Next9 consider
o , ,
f (n>k>r) =
In - ka + a j
Jr
Since f N(n.k.r) is an r-nomial coefficients then

a
f (n + l , k , r ) = f (n,k,r) + f (n - a, k - l , r ) + + f (n - sa 9 k - s s r)
a
a
a
a
+ . . . + f (n - (r - l)a, k - r + l , r ) .
a
460
[Dec.
Thus, f (n,k,r) and g (n,k,r) both obey the same recurrence relation, and
fa(n,l,r)
= ga(n,l,r) = n
f ( l , n , r ) = g ( l , n , r ) = 0,
d
n > 1
d.
for all n > 0, so that the arrays are identical for all k > 0.
Summing, for fixed n > 0, over all numbers of all k-part subsets
yields
Rn+a)(r-l)1
L l+a(r-irj
T7 /
V*
J n - ka + a [
Va(n,a,r) =
fc
If we now generalize the "generalized Fibonacci numbers,
u(n; p,q), of
Tf
Harris and Styles [3] to the generalized Pascal triangles obtained from the
expansions (1 + x + x2 + + x r " 1 ) , n = 0, 1, 2, 3, ,
f n(r-l) 1
[q+p(r-l)J
u(n;p9q,r) =
"qk^
k=0
there are precisely
ones at the beginning of each u(n; p , q , r ) sequence.

with just one 1. Let
[A]
the greatest integer in q/(r - 1). Then,
Our application starts
1970]

Rn+a+mHr-lfl
L b+a(r-l) J
u(n + a + m; a , b , r )
461
n + a + m - kal
kb
|
k=0
Thus the solution set to the number of subsets of {l, 2, 3, a , n} subject

to the constraints that no pairs i,
i +j
(j = 1, 2, 3, 5 a)
are to be
allowed inthe same subset, nor are all of i + ja + 1 (j = 0 , 1 , 2 , 3 , - * - , r - 1)

to be allowed in the same subset, are the generalized Fibonacci numbers of
Harris and Styles generalized to Pascal triangles induced from the expansions
of
(1 + x + x2 + + x
-1
) ,
n = 0, 1, 25 3S " .
One notes that the r-nacci generalized Fibonacci numbers
rn(r-l)'
u(n; 1,1, r) =
Jj
| n - kj
k=0
are not generally obtained by setting a = 0 in the above formulation.
How-
ever, the generalized Fibonacci sequences for the binomial triangle are obtained if r = 2e The other r-nacci number sequences are obtained if the
subsets are simply restricted from containing simultaneously r consecutive
integers from the set {l, 2, 3, * * - , n} but there is no restriction of r > 2
about pairs of consecutive integers 8
Thus, for these
r-nacci
sequences
(r > 2), we cannot simply set a = 1. However, the formulas look identical.
Let
V(n; 1,1, r) = u(n + 1; 1,1, r) ;
then

462
Dec. 1970
(n+l)(r-l)1
r
V(n; 1,1, r) =
^
k=0
n - k + ll
k
which is seen to be the generalization of Kaplansky f s lemma to generalized

Pascal triangles.
REFERENCES
la John Riordan, Combinatorial Analysis^ Wiley and Sons, Inc, New York,
p. 198.
2. John Riordan, Combinatorial Analysis, Wiley and Sons, Inc., New York,
p. 222.
3. V. C. Harris and Carolyn C. Styles, !fA Generalization of Fibonacci
Numbers," Fibonacci Quarterly, 2 (1964), pp. 277-289.
4. D. E. Daykin, "Representations of Natural Numbers as Sums of Generalized Fibonacci Numbers," Journal of the London Mathematical Society,
April 1960, pp. 143-168.
5. S. Go Mohanty, "On a Partition of Generalized Fibonacci Numbers,"
Fibonacci Quarterly, February 1968, pp. 22-33.
6. V. E. Hoggatt, J r ,
"A New Angle in PascaTs Triangle," Fibonacci
Quarterly, Oct. , 1968, pp. 221-234.

7. Nannette Cox, John W. Phillips, and V. E. Hoggatt, J r . , "Some Universal Counterexamples," Fibonacci Quarterly, April, 1970, pp. 242248.
8. C. A. Church, J r . , "On Enumeration of Certain Triangular A r r a y s , "
Fibonacci Quarterly, April, 1970, pp. 235-241.
9. J. E. Freund, "Restricted Occupancy Theory A Generalization of
Pascal's Triangle," American Mathematical Monthly, January, 1956,
pp. 20-27.
10. C. A. Church, J r . , "Combination and Successions," Mathematics Magazine, May-June, 1968, pp 123-128.
11. Nannette S. Cox, "Unique Integer Representation by Generalized Fibonacci Numbers Derived from Pascal's Triangle," SanJose State College
Masters Thesis, July, 1970.
12. John W. Philips, "Integer Representation as Sums of Generalized Fibonacci Numbers," San Jose State College Masters Thesis, June, 1970.
APPLICATION OF RECURSIVE SEQUENCES

TO OiOPHANTINE EQUATIONS
RAPHAEL FINKEISTEIN
Bowling Green State University, Bowling Green, Ohio
EDGAR KARST
University of Arizona, Tuscon, Arizona
HYMIE LONDON
McGill University, Montreal, Canada
ABSTRACT
In a f o r m e r v e r s i o n of t h i s p a p e r ("Iteration A l g o r i t h m s for C e r t a i n
Sums of S q u a r e s " ) 5 K a r s t s by composition of simple s u m s of s q u a r e s 9
found
six i t e r a t i o n a l g o r i t h m s of which he could p r o v e the f i r s t by m e a n s of the gene r a l i z e d P e l l equation and the second by the p e r m a n e n c e of f o r m a l laws*
the r e m a i n i n g four, equivalent to the solution of x 2 - (k2 + l)y 2 = k 2 ?
For
with k
= . 1 , 2 , and 3 5 F i n k e l s t e i n and London w e r e able to furnish a unifying proof

by the u s e of c l a s s n u m b e r s and q u a d r a t i c fields* T h i s justifies the new title.
The following t h r e e - s t e p i t e r a t i o n a l g o r i t h m to g e n e r a t e x in 2x + 1 =
a
and 3x + 1 = b 2 9
s i m u l t a n e o u s l y ! w a s mentioned in [6, p. 2 1 1 ] :
9 2 = 81
1-10-1 = 9
9-10-1 = 89
89
89-10-9 = 881
= 7921
= 776161
= 765055841
881
881-10-89 = 8721 8721
( 8 1 - D / 2 = 40 =
Xl
(7921-1)/2 = 3960 = x 2
(776161-1)/2 = 388080 = x 3
(76055841-1)/2 = 38027920 = x 4
8721- 10-881 = 86329 86329 2 =7452696241 (7452696241-D/2 = 3726348120 = x 5

Proof.
a
From
2x + 1 = a2
and 3x + 1 = b 2 c o m e s 3a 2 - 2b 2 = 1.
is any solution of this g e n e r a l i z e d P e l l equation, then

a ,- = 5a + 4b ,
n+1
n
n
i s the next l a r g e r one
b ^ = 6a + 5b
n+1
n
n
F r o m t h e s e , we can obtain i m m e d i a t e l y
+ a ., = 10a ,
n+1
n-1
n
b ,- + b - = 10b ,
n+1
n-1
n
which i s equivalent to the algorithm,,

463
If
464
APPLICATION O F RECURSIVE SEQUENCES

F o r the n
[Dec.
f o r m u l a , we u s e the usual a p p r o a c h by l i n e a r substitution
(for e x a m p l e , [ 1 , p. 181]) and obtain

x
[<V6 + 2)(5 + 2 \ / 6 ) n + (\/6 - 2)(5 - 2 \ / 6 ) n ] / 4 8 - 1/2
T h i s formula h a s t h r e e s h o r t c o m i n g s :
(1) it u s e s f r a c t i o n s ,
r o o t s , ancl (3) it h a s n in the exponent.
(2) it employs
The a l g o r i t h m above h a s none of
them.
S i m i l a r a r g u m e n t s a r e valid for a f o u r - s t e p i t e r a t i o n a l g o r i t h m [4] to
g e n e r a t e x in x 2 + (x + I) 2 = y 2 .
t e r m f o r m u l a m a y be s i m p l e , .as for a2 + b 2 + (ab) 2
S o m e t i m e s , the n
= c2,
a and b consecutive positive i n t e g e r s [ 2 ] .

(n - I) 2 + n 2 +
(n - l)n
and hence we need no a l g o r i t h m .

helpful.
But for
H e r e , we have
= (n2 - n + I) 2 ,
a = 1,
an a l g o r i t h m would be
L e t u s f i r s t find s o m e c l u e s to such an a l g o r i t h m . We have by hand
and by a table of s q u a r e s :
12 +
o2 +
02 =
l2 =
(02 + I ) 2
l2 +
22 +
22 =
32 =
(22 - l ) 2
I 2 + 12 2 + 12 2 = 17 2 =
(42 + l ) 2
+ 70 + 70
= 99
= (102 - l ) 2
The a l t e r n a t i n g +1 and - 1 in the l a s t column, which shows a constant p a t t e r n , s u g g e s t s the possibility of an a l g o r i t h m .

from
If we can find all b ,
say
D3 = 12 on, we will also have all c. After s o m e t r i a l s and e r r o r s we
obtain
Iteration Algorithm I
6-2 - 0
6-12 - 2
6-70 - 12
6-408 - 70
6-2378 - 408
6-13860 - 2378
=
=
=
=
=
=
12
70
408
2378
13860
80782
1970]
TO DIOPHANTINE EQUATIONS
465
which yields e a s i l y the next four r e s u l t s :

I2 +
408 2 +
408 2 =
577 2 =
(242 + I) 2
l2 +
2378 2 +
2378 2 =
3363 2 =
(58 2 - I) 2
l 2 + 13860 2 + 13860 2 =
19601 2 = (140 2 + I) 2
I 2 + 80782 2 + 80782 2 = 114243 2 = (338 2 - I ) 2

S i m i l a r l y s we approach the c a s e a = 2 ,
We have by hand and by a table of
squares:
22 +
I2 +
22 =
3 2 = (I 2 + 2) 2
22 +
32 +
62 =
72 = (3 2 - 2) 2
22 +
82 + 16 2 = 18 2 = (42 + 2) 2
2 2 + 2 1 2 + 42 2 = 47 2 = (72 - 2) 2
The a l t e r n a t i n g +2 and - 2 in the l a s t column, which shows a constant p a t t e r n , suggest the p o s s i b i l i t y of an a l g o r i t h m . If we can find all b , say from
b 3 = 8 on, we will also have all c After s o m e t r i a l s and e r r o r s we obtain
I t e r a t i o n Algorithm II
3-3 - 1 = 8
3-8 - 3 = 21
3-21 - 8 = 55
3-55 - 21 = 144
3*144 - 55 = 377
3-377 - 144 = 987
which y i e l d s e a s i l y the next four r e s u l t s :

22 +
2
55 2 +
HO 2 =
123 2 = ( l l 2 + 2) 2
322 2 = (18 2 - 2) 2
754 2 =
843 2 = (292 + 2) 2
+ 144
2 2 + 377 2 +
288
2 2 + 987 2 + 1974 2 = 2207 2 = (472 - 2) 2 .

Slightly differently behaves the c a s e of a = 3*
of s q u a r e s
We have by hand and by a table
466
[Dec.

32 +
02 +
02 =
32 +
62 =
32 +
22 +
42 +
32 +
18 2 +
542 =
32 +
80 2 +
240 2 =
2
253 2 = (16 - 3)2
3 2 + 1542 + 4622 =
4872 = (222 + 3)2
12
(02 + 3)2
32 =
72 =
13
(2 2 + 3)2
(42 - 3)2
57 2
32 + 6842 + 20522 = 21632

Here the doubly alternating +3 and -3 in the last column would show a constant pattern, if the exceptional values 572 and 21632 could be eliminated.
This suggests obviously the possibility of two algorithms. To obtain further
results, we write an Integer-FORTRAN program for the IBM 1130 which
yields:
32 +
3038 2 +
9114 2 =
9607 2 =
(982 + 3) 2
32 +
5848 2 +
17544 2 =
18493 2 =
(136 2 - 3) 2
32 +
25974 2 +
77922 2 ="
82137 2
32 +
115364 2 +
32 +
222070 2 +
32 +
986S28 2 +
346092 2 =
666210
2958984 2 =
364813 2 =
(6042 - 3) 2
702247 2 =
(838 2 + 3) 2
3119043 2
3 2 + 4380794 2 + 13142382 2 = 13853287 2 = (3722 2 + 3) 2
Now we want to find an algorithm which should generate the sequence 80, 154,
3038, 5848, 115364, 222070, 4380794, .
and b 3 = 4 be given;
32
Let the terms bt = 0,
b2 =
2,
then b 0 = -4 is the left neighbor of ht = 0, since
+ (_ 4 )2 + ( 1 2 ) 2
is. the logical extension to the left.
132
( 4 2 3)2
With this new initializing and some trials
and e r r o r s , we obtain the Iteration Algorithm III on the following page.
Now
there remains only to find an algorithm which should generate 25974, 986328,
e . Here, we have not far to go, since such an algorithm is already contained in the former one, and we obtain Iteration Algorithm IV on the following
page.
1970]
TO DIOPHANTME EQUATIONS
467
Iteration Algorithm HI
38*2 - (-4)
280 - 2:4 + 2
38-80 - 2
2-3038 - 2*154 + 80
38*3038 - 80
2*115364 - 2t5848 + 3038
38-115364 - 3038
=
=
=
=
=
=
=
80
154
3038
5848
115364
222070
4380794
I t e r a t i o n Algorithm IV
38-684 - 18 = 25974
38*25974 - 684 = 986328
F i n a l l y s one could ask: Does t h e r e exist a g e n e r a l formula for solving x 2 +
y 2 + z 2 = w2 ? The a n s w e r i s yes 0
x = p 2 4- q2 - r 2 ,
then
y = 2pr 3
x 2 + y2 + z 2 = w2
comingss
Let
becomes
z = 2qr 5
0 = 0.
w = p 2 + q2 + r 2 ;
and
But this f o r m u l a h a s two
short-
(a) it u s e s f r a c t i o n s , and (b) it employs r o o t s , s i n c e , for e x a m p l e ,
the solution
3 2 + 2 2 + 62 = 72
requires
p =\/~2/29
q = 3\/2/2s
and
r =
V2.
Now we shall p r o v e how the i n t e g e r solutions of c e r t a i n Diophantine
equations of the second d e g r e e , equivalent to Iteration A l g o r i t h m s I-IV,
be found by r e c u r s i v e s e q u e n c e s .
can
We will c o n s i d e r the equation
x 2 - (k2 + l)y 2 = k2
(1)
with k = 1, 2 9 and 3.
Further, x
and y
will denote i n t e g e r s o l u t i o n s of
(1).
If k = 1,
Eq. (1) b e c o m e s x 2 - 2y 2 = 1.
By T h e o r e m 3 of [ 3 ] , the
r e c u r r e n c e formula for this equation i s given by
y
J
n
= 6y ,- - y - ,
J
J
n+1
n-1
with y1 * 2 and y2 = 12.
n > 2
468

If k = 2 ,
[Dec.
Eq. (1) b e c o m e s
x 2 - 5y 2 = 4 .
(2)
T h i s equation belongs to the q u a d r a t i c field
Q(0),
0 = V^s
which h a s
(1, (1 + 0 ) / 2 ) a s an i n t e g r a l b a s i s , and i t s fundamental unit i s 0 = (1 + 0)/2,,

Since the c l a s s n u m b e r of
Q(0)
i s 1 and the d i s c r i m i n a n t D = 5 (mod 8),
the ideal (2) i s p r i m e [5, p. 6 6 ] ,
H e n c e , all the a l g e b r a i c i n t e g e r s of Q(0)
of n o r m 4 a r e a s s o c i a t e s of 2.
T h u s , if x
+ y 0 is an a l g e b r a i c i n t e g e r of
n o r m 4 , we get
x
o 2n
n
+ Jy 0n = 2u0
= n26-t
1
n
9,
w h e r e et = e\ = (3 + 0 ) / 2 .
Remark.
Since we want all the a l g e b r a i c i n t e g e r s of n o r m 4 , we have

only c o n s i d e r e d the even p o w e r s of 0 .
n + l
_ o n
Noting that
n-1
we obtain
yJ , i
n+1
3y
- Jy - ,
J
n
n-1
n > 2 ,
with ji = 1 and y 2 = 3. It can e a s i l y be shown, by using the

well-known identity
L 2 - 5F 2 = 4N( - l ) n
n
n
of the L u c a s and Fibonacci n u m b e r s , that y
= FQ
n
= L? .
(3)
If k = 3 , Eq. (1) b e c o m e s
x2 -
10y 2
= 9 .
T h i s equation belongs to the field Q(0), 0 = \ / T 0 ,

integral basis,
and x
6X1
which h a s
(1,0) a s an
0 = 3 + 6 a s i t s fundamental unit, and c l a s s n u m b e r
2.
1970]
TO DIOPHANTME EQUATIONS
469
Since the discriminant D = 1 (mod 3), the ideal (3) becomes PjPg, where
Pi and P 2 are distinct prime ideals of norm. 3. Thus there are 3 distinct
ideals of norm 9. Since 3, 7 - 26, 7 + 26 are non-associated integers of
norm 9, all the integers of norm 9 are associates of one of these 3 integers.
It follows that
X
(4)
3n
Jan*
3 n
3n+1
(7
3n+2
yan+2 6
( 7 + W)
'
"
26 R
= <7 - 2*>?
(7 +
W)
By applying Theorem 3 of [ 3 ] , we find that e? satisfies the recurrence

formula
u , 0 = 38u (1 - u
,
n+2
n+1
n '
where u is either the constant term of the coefficient of 6 for ^j\
the recurrence formulas for Eqs. (4) a r e , for n > 2,
bQ
= 38b Q Q
3n
3n-3 "
b Q ^ = 38b Q 0
3n+l
3n-2 "
b
3n+2
38b
3n-l -
3n-69
3n-5>
3n-4>
= 57,
bj = 2,
b2 = 4 ,
Thus
b 6 = 684,
b 4 = 80,
b 5 = 154
REFERENCES
1* Irving Adler, "Three Diophantine Equations, n Fibonacci Quarterly, 6

(1968), pp0 360-369; 7 (1969), pp0 181-193.
2. American Mathematical Monthly, 76 (1969), p. 187.
3. E. L Emerson, "Recurrent Sequences in the Equation DQ2 = R2 + N,
Fibonacci Quarterly, 7 (1969), pp. 231-242.
4. Edgar Karst, ?!A Four-Step Iteration Algorithm to Generate x in x2 +
(x + I) 2 = y 2 , f ! Fibonacci Quarterly, 7 (1969), p. 180.
5. H. B. Mann, Introduction to the Theory of Algebraic Numbers, The Ohio
State University P r e s s , Columbus, Ohio, 1956.
6. Zentralblatt fiir Mathematik, 151 (1968), pp. 211-212.
ON A CLASS OF DIFFERENCE EQUATIONS

RAYMOND E. WHITNEY
Lock Haven State College, Lock Haven, Pennsylvania
The purpose of this a r t i c l e i s to examine s e q u e n c e s g e n e r a t e d by a c e r tain c l a s s of difference equations and to encourage further investigations into
their properties.
We shall be i n t e r e s t e d in s e q u e n c e s satisfying the r e c u r -
rence relation,
(1)
v _, 0 = v , 1 + v + kv v , n ;
n+2
n+1
n
n n+1
vi = Vo = 1
\
L
(n > 1) ,
/
w h e r e k i s a positive i n t e g e r .
It m a y be shown by a s i m p l e inductive a r g u m e n t that
F
(2)
vn =
~-
(n ^ 1) ,
where F
denotes the n
Fibonacci n u m b e r .
n
When we wish to e m p h a s i z e the dependence on the p a r a m e t e r ,
shall w r i t e v
k,
we
5 v (k).
n
A MODEL FOR (Lv J} ,
n n=l
Let b denote an i n t e g e r
(b > 2).
C o n s i d e r the sequence defined a s
follows:
F
(3)
where
n
6Q = 1 1 . . - 1
(b) denotes b a s e b.
6R = E
i=0
(n > 1)
Obviously,
F -1
(4)
(b)
F
1
b = ^ ^ i
D
470
(n> 1)
Dec. 1970
As above* we shall write 6n = 6R(b),
471
From Eqs. (2) and (4), we see that
v n (k + D = ena>) .
b - 1 has been called the n
Fermatian function of b and
n = T^T
has been called a reduced Fermatian of index be

F
(See [1].)
We note that
n
n
If we are willing to abuse the language, we may extend the allowed
values of b. Formally 9 if k ~ o s Eqa (1) becomes the usual Fibonacci r e currence relation. Then b = k + 1 = 1, and if we interpret the l f s in (3) as
tally marks,
F -1
n
en = i d )
+ . . . +i(i)
F
= 11 . . . I
(1) .
Similarlyj if k = - 1 , then b = 0e With the agreement that 0 = 1,

F -1
en = i(o)
+ . . . + KO)0
= 1 1 - . . 1 (0)
Thus 6 = 1. But the solution of (1) in this case is
v n (-l) s i
(n > 1) .
Using similar interpretations for negative bases, we can extend (1) and (3) to
negative integers*
472
[Dec.
DIVISIBILITY PROPERTIES OF { v j ^
It is interesting to note that if
contains an infinite number of primes, then there would be an infinite number

of Fibonacci and Mersenne primes.
In this section, we shall assume k = 9 (b = 10) unless otherwise
specified.
Theorem 1.
(a)
(6 n ,
n+1)
(b)
<Vn+2)
= 1
(n> 1);
(n
^1)-
Proof, a) Deny! Then there is a pair such that (0 ,0 - ) = d > 1.

But d|v + 2 , d|v - implies d|v . Thus, after repeated use of the above,
we would have (0i,0 2 ) ^ l e Contradiction.
b) Similar to part a).
Theorem 2. None of the 0 are perfect.
n
Proof. Any odd perfect number
is^ congruent to 1 modulo 4 (see [2]).
But
0 = 3
Theorem 3. 3|fl
Proof.
110
i n
for
n > 3,
if and only if 4|n.
Proof. Clearly, s\e

suit follows.
Theorem 4. 11 \6
(mod 4)
if and only if S | F Thus F J F
and t h e r e -
if and only if 3|n.
if and only
J if 2 = F 0 1 F
3
and the result follows.
Theorem 5. a) 7 0 if and only if 12 n;

b) 1310 if and only if 12 |n.
Proof,
a) Consider the congruences,
1 = 1 (mod 7) ,
10 = 3 (mod 7) ,
1,000 = - 1 (mod 7),
10,000 = - 3 (mod 7),
100 = 2 (mod 7),

100,000 = -2 (mod 7).
1970]
473
Clearly 7|fl
if and only if 6 | F .
and
But 6 |F
is equivalent to 2 | F
of 3|n and 4|n and the result follows,

b) Similar to a) 9 considering the congruences modulo 13.
In light of the above, we have the unusual result that 3\6 and l l | e
implies T| 6> and 13|# .
*
' n
' n
We mention some other results which the reader might like to establish.
Assertion Is
18 |F
Assertion 2:
Assertion 3;
41
Assertion 4s
implies 19|0 .
l ^ n ^ anc^ o n ^ y ^ 5 l n
271|0 if and only if 5|n.
73J0 , 10l|6 ,
137J0
if and only if 6|n .
GENERATING FUNCTIONS FOR {v ( k ) } * ^

One area which might be worth investigating is that of obtaining generating functions for the sequences. Of course, since
(6)
i-i
1 - x - x2
=Z V
i=l
we have
oo i o g [ l + kv.(k)]
(7)
. ,
77777-5 * * *
but one should be able to do better than this.

ALTERNATE RELATIONSHIPS
We present two results along these lines.
Theorem 6.
6
Proof.
(2) = 2 TT [1 + 0.(2)] - 1
i=l
Since
F
2
= i + e (2)
(n > 1)
474
[Dec.
and
1=1
i = F n + 2 -1
(n
" X)
the result easily follows.,

Theorem 7.
i + e 2n (2) = 7 T [i + 0 2 i _i ( 2 ) 1
Proof
< n ^ D-
The result is readily obtained from
j F2i-1
= F
1=1
2n
GENERALIZATION TO OTHER RECURSIVELY DEFINED SEQUENCES

We conclude our discussion with one result in this area.,
Theorem 80 If
1
nJ n=l
is a recursively defined positive integer sequence satisfying the linear difference equation
(8)
m
û ^ = P
i=0 * n + 1
(n > 1)
(order m)
and boundary conditions {u l9 u 2 , B, u m - l ) where jS and a. for

1, , m} are constants, and if
i(0,
1970]

u
n
= 1 1 - . 1 (b)
0n
475
(n > 1) ;
then
ni
a.
0 [1 + (b - l)p
] i = b13
n+1
i=0
(9)
Proof,
(n > 1)
Since
u
^n = T A H T
<n * x>
we have
u
k
b ~ = 1 + (b - 1)/^
for k > 1 and the result readily follows,

REFERENCES
1 L. E. Dickson, History of the Theory of Numbers, VoL 1, pa 3858
2. Ibid, p. 19,
ON A CONJECTURE OF DMITRI THORO*

DAVID G. BEVERAGE
Sao Diego State College, San Diego, California
Denoting the n
term of the Fibonacci sequence 1, 1, 2, 3, 5, ,
byJ F n9, where F n+2

, 0 = Fn+1
, - + Fn , it is well known that
F2 - F
F
= (-l)n+1
K 1;
n
n-ln+1
If odd prime p divides F
- , then
F2 = ( _ D n + 1
so that (-1)
(modp) ,
is a quadratic residue modulo p* Clearly, for n = 2k, this
implies - 1 is a quadratic residue modulo p , and accordingly, p = 1 (mod

[Continued on page 537* ]
THE FIBONACCI NUMBERS

CONSIDERED AS A PISOT SEQUENCE*
MORRIS JACK DeLEON
Florida Atlantic University, Boca Raton, Florida
C h a r l e s P i s o t [1] w a s the f i r s t to c o n s i d e r the s e q u e n c e , { a n } 0 0 _ o S
of
n a t u r a l n u m b e r s d e t e r m i n e d from two n a t u r a l n u m b e r s a 0 and aj such that

2 < a0 < a4
(1)
and
a2
I <a
"2
for all n > 0.
n+2
n+1
<A
a
- 2
n
The Fibonacci n u m b e r s with the f i r s t two t e r m s deleted s a t -
isfy Eq. (1).

P e t e r F l o r [2] called the s e q u e n c e s which satisfy (1) P i s o t s e q u e n c e s
of the second kind.
F l o r a l s o c o n s i d e r e d the sequence of n a t u r a l n u m b e r s
d e t e r m i n e d from two n a t u r a l n u m b e r s a0 and aj such that

2 < a0
(2)
< at
and
A<
2 -
n+2
_!*+!<
I2
a
n
for all n > 0.
He called t h e s e s e q u e n c e s P i s o t s e q u e n c e s of the f i r s t kind.
F o r a P i s o t sequence of the f i r s t (second) kind a

i n t e g e r to
-/a
i s simply the n e a r e s t
, w h e r e in c a s e of ambiguity we choose the
smaller
* T h i s p a p e r i s taken from C h a p t e r F o u r of the author 7 s d o c t o r a l d i s s e r t a t i o n

a t the Pennsylvania State University. The author would like to e x p r e s s h i s
gratitude to R e s e a r c h P r o f e s s o r S. Chowla* his a d v i s o r .
476

1Q7n
477
u e c . iy/u
CONSIDERED AS A PISOT SEQUENCE
(larger) integer. By Pisot sequence we shall mean a sequence that satisfies
(1) and (2).

By
{F
L
n
+ kJ fn=n
we mean the sequence formed by adding k to each term of the sequence
Kr .
where F
is the n
n=n0
Fibonacci number. In this paper necessary and suf-
ficient conditions for
{F
+k}
1
J
n
n==n0
to be a Pisot sequence are given,

The main result is
Theorem.
Let
1
nJ n=l
be the Fibonacci sequence. The sequence

{F
1
J
n + l}
n=n0
is a Pisot sequence of the first kind (second kind) iff n0 > 6 (n0 > 4),
The
sequence
(1 F n - l}
J
n=]
n=n0
is a Pisot sequence iff n 0 > 7. The sequence {^n}
is a Pisot sequence of
the first kind (second kind) iff n 0 > 4 (n0 > 3). If |k| > 1 then there exists
no integer n0 such that
(F
1
+ k}
}
n=n0
478
[Dec.
is a Pisot sequence.
We shall need two l e m m a s in o r d e r to p r o v e the t h e o r e m .
L e m m a 1.
F x 0 - 2 F M +F
= F 0 .
n+2
n+1
n
n-2
Proof. F ^ Q - 2F _ + F = (F _ + F ) - 2 F ^ + F
n+2
n+1
n
n+1
n
n+1
n
= - F ^ + 2 F = - ( F + F n ) + 2F
n+1
n
n
n-1
n
= F
L e m m a 2.
= ^
Proof.
n " Fn-1 = Fn-2 "

+ k) - (FR+1 + k) 2
( F n + 2 + k)(FR
(F
JQ
n+2
(r
+ k)(F
::
+ kF
n-2 '
+ k) - (F , . + k) 2
n
n+1
r - Fn+1>
+ k(
V2 -
2F
The l a s t equality i s t r u e since

2
/ ixn+1
F ^ 0 F - F^
= (-1)
n+2 n
n+1
We a r e now able to begin the proof of the t h e o r e m .
F r o m the definition
of a P i s o t sequence and L e m m a 2 , we have that
n=n 0
i s a P i s o t sequence of the f i r s t kind iff

(i)
2 < F
+ k < F
n0
^ + k
n 0 +l
and
(iia)
(iib)
- ( F n + k)
2 [ ( - l ) n + 1 + kF M
a r e satisfied.
(i) and
< 2[(-l)n+1 + kFn_2]
] < FM + k
for all n > n 0
for all n > n 0
Also (1 F + k )J
i s a P i s o t sequence of the second kind iff
n
n=no
1970]
CONSIDERED AS A PISOT SEQUENCE

n+1
(iiia)
-(F n + k) < 2 [ ( - l )
+ kF n _ 2 ]
(iiib)
2[ ( - l ) n + 1 + kF n _ 2 ] < F n + k
479
for all n > n0

for all n > n0 .
We shall first consider the case k = 1.

F
+ 1
2F
+ F
n-2
n-3
+ 1
>
2 F
[ n-2
^ *
2 F
[ n-2
iff n > 6. Thus (iib) is satisfied iff n > 6* Also,

F
+ 1 = 2F
n-20
+ F
Q
n-3
+ l > 2[F
L
0
n-2
+ 1]J > 2[F

L
n-20
+ ( - l ) n + 1 J]
iff n > 4. Thus (iiib) is satisfied iff n > 4, Since

2[(-l)n+1 + F
> 0 > -(F
+1)
for all n > 3,
(iia) and (iiia) are satisfied for n > 3 It is clear that (i) is satisfied if n0 >
2 Thus
0
(1F n + l }J 0n=n
0
is a Pisot sequence of the first kind iff n 0 > 6 and it is a Pisot sequence of
the second kind iff n 0 > 4
Next s we consider the case k = - 1 . If n = 6 both (iia) and (iiia) are
not satisfied.
F
If n = 7, both (iia) and (iiia) are satisfied.,
- l = 2F
n
n-2
+ F
- 1 > 2[F
n-3
+ 1]
J
n-2
Now
if n > 8
Thus,
-(Fn-1)
< 2[-l-Fn_2]<
2[(-l)n+1-Fn_2]
if n > 8 Therefore, (iia) and (iiia) are satisfied iff n > 7* Since
480

2 L[ ( - l )
n+1
- F
[Dec,
J < 0 < F
-1
n-2 J
n
if n > 3, both (iib) and (iiib) are satisfied for n > 3, It is clear that (i) is
satisfied for n > 4. Thus,
{F
* n
- i} 0 0
J
n=n0
is a Pisot sequence iff n 0 > 7.

Now we consider the case k = 0,
It is clear that (i) is satisfied iff
n > 3. Both (iia) and (iiia) are satisfied for n > 3. Also (iiib) is satisfied
for n > 3, but (iib) is satisfied iff n > 4, Thus
{F
}
n=n0
is a Pisot sequence of the first kind iff n 0 > 4, and
(F }
n=n0
is a Pisot sequence of the second kind iff n0 > 3.
We shall show that if |k| > 1, then there exists no integer n0 such
that
{F *k}
is a Pisot sequence. This will be accomplished by showing that (iia) or (iiia)
implies that k > -2 and that (iib) or (iiib) implies that 2 < k.
Dividing (iia) by F 0 yields that
ndi
F
F
0/
^vn+1
- . J . ^JS- < 2^1)

F
n-1
n-2
n-2
+ 2k
n-2
for n > n0 After taking the limit of both sides as n oo and remembering
that
1970]
CONSIDERED ASA PISOT SEQUENCE
481
limVi=Ll_vi<2 >
n
we have that
-M
-4 < - l i m I * 1 < 2k
Thus
-2 < k .
In a similar manner, one can show that (iib) or (iiib) implies that k <
2.
REFERENCES
1* Charles Pisot, "La Repartition modulo un et les Nombres Algebriques,"
Ann, Scuola Norm. Sup. Pisa (2) 7 (1938a)9 pp. 205-248,
2, Peter Flor, "Uber eine Klasse von Folgen Nattirlicher Zahlen," Math.
Ann. 140 (I960), pp. 299-307.
FALL RESEARCH CONFERENCE

St. Mary's College, Saturday, October 17, 1970
9:15
Registration
10:0010:50 Combinatorial Problems Leading to Generalized Fibonacci
Numbers. Verner E. Hoggatt, J r . , San Jose State College
11:0011:50 How Fibonacci Numbers Helped Solve Hilbert ! s Tenth Problem
Professor Julia Robinson, University of California, Berkeley
1:302:20 Explicit Determination of Perron Matrices. Professor Helmut
Hasse, Visiting Lecturer, San Diego State College
2:303:20 Asymptotic Fibonacci Ratios. Brother Alfred Brousseau, St.
Mary's College
3:304:00 Fibonacci Correlations in Bishop Pine. Brother Alfred
Brousseau, St. Mary's College.
MODULO ONE UNIFORM DISTRIBUTION OF THE SEQUENCE

OF LOGARITHMS OF CERTAIN RECURSIVE SEQUENCES
J. L. BROWN, JR., and R. L. DUNCAN
The Pennsylvania State University, University Park, Pennsylvania
Let {x.} be a sequence of real numbers with corresponding fractional

parts {/3.}, where 0. = x. - [x.] and the bracket denotes the greatest integer function.
For each n > 1, we define the function F
on [ 0,1] so that
F (x) is the number of those terms among /31$ , /3R whichlie in the interval [0,x), divided by n. Then {x.}
is said to be uniformly distributed
modulo one iff nlim oo*n

F (x) = x for all x TO,
In other words, each
L 11.
J
interval of the form [0,x) with x [ 0 , 1 ] , contains asymptotically a proportion of the p f s equal to the length of the interval, and clearly the same
will be true for any subinterval (a9p) of [0,1].
ion ([1], p. 76) states that {x.}
00
is uniformly distributed mod 1 iff
lim i Y e2nivXj
n oon i
(1)
The classical Weylcriter-
= 0
> 1.
j=l
An example of a sequence which is uniformly distributed mod 1 is { nf} ( 1 0

where is an arbitrary irrational number (see [1], p. 81 for a proof using
Weyl's criterion).
The purpose of this paper is to show that the sequence (in V } is uni" l
formly distributed mod 1, where ( v } is defined by a linear recurrence of

the form
(2)
V , = a, ,V , , + +a A V
n+k
k-1 n+k-1
On
the initial terms Vj, V2s
n > 1,
8 9
, V. being given positive numbers. In (2), we
assume that the coefficients are non-negative rational numbers with a0 ^ 0,

k
k-1
and that the associated polynomial x - a, -x
- - a-jx - a0, has roots
eo
Pi> ft ? Pk which satisfy the inequality 0 < \Pi\< < j/3, |. Additionally 9 we require that |/3.| f 1 for j = 1, 2, , k.-.
482
D e c . 1970
MODULO ONE UNIFORM DISTRIBUTION
483
In p a r t i c u l a r s o u r r e s u l t i m p l i e s that any sequence
{u } which s a t n J
isfies" the Fibonacci r e c u r r e n c e U , n = U M + U
for n > 1 with U< = k-i
n+2
n+1
n
~
i
i
and U2 = k 2 a r b i t r a r y positive initial t e r m s (not n e c e s s a r i l y i n t e g e r s ) will
have the p r o p e r t y that ( i n U } i s uniformly d i s t r i b u t e d m o d 1. With kA =
1, k 2 = 1, we obtain this conclusion for the c l a s s i c a l Fibonacci sequence
(see [ 2 ] , T h e o r e m 1), while for
k 1 = 1,
k2 = 3 ,
an analogous r e s u l t is
s e e n to hold for the L u c a s sequence.

Before proving the m a i n t h e o r e m , we p r o v e two l e m m a s :
L e m m a 1.
If ( x . )
that
.lim
i s uniformly d i s t r i b u t e d m o d 1 and {y.}
J i
( x . - y . ) = 0,
Proof.
then \ y . )
i s such
J i
.oo
is uniformly d i s t r i b u t e d mod 1.
F r o m the hypothesis and the continuity of the exponential func-
tion, it follows that

lim
(e
1 - e
j) = 0 .
But it i s well known ( [ 3 ] , T h e o r e m B , p. 202) that if {y } is a sequence of

r e a l n u m b e r s converging to a finite l i m i t L ,
then
n
> y.
lim
n 00
2>i ==
1
rr, , .
27rii/x.
2irii/y.
u
J
Taking y. = e
j - e
] , we have
1 \ ^
t.
lim - >
n - 00 n Z-^
, 2-nivx.
2mvy.,
J
(e
] - e
j)
Since
n
1
\"^
lim 7 e
a-+oo n Zi
1
by Weyl's c r i t e r i o n , we also have
27T1VX.
1 = 0
= 0
484
[Dec.
n
27ri y
Ibn
i
V
n ~* n ^-^ e ^ Jj = 0
and the sufficiency of Weyl's criterion proves the sequence {y.} to be uniformly distributed mod 1.
Lemma 2. If a is a positive algebraic number not equal to one, then
In a is irrational.
Proof. Assume, to the contrary, In a = (p/q), where p and q are
non-zero integers. Then e p / q = a9 so that e p = cfl. But or is algebraic,
since the algebraic numbers are closed under multiplication ([1], p. 84).
Thus e p is algebraic, in turn implying e is algebraic. But e is known to
be transcendental ([1], p. 25) so that a contradiction is obtained.
Theorem. Let \ V } be a sequence generated by the recursion relation,
(2)
V M = a. nV M - + - + a i v _LI
n+k
k-1 n+k-1
1 n+1
+ a
nV
On
(n > 1) ,
where a 0 , a l s , a^__^ are non-negative rational coefficients with a0 ^ 0,

k is a fixed integer, and
(3)
Vi = yi9
v2 = y2, ,
v k = yk
are given positive values for the initial terms. Further, we assume that the
k
k-1
polynomial x - ab- i x
- - a^x - a 0 has k distinct roots j3l9 /32, ,
ft satisfying 0 < \pt\ < < j ^ | and such that none of the roots has magnitude equal to 1. Then (in V } is uniformly distributed mod 1.
Proof. The general solution of the recurrence (2) is
k
(4)
Vn=5>j<?
(n>l),
j=l
where the arbitrary constants al9 a2, , a^ are determined by the specification of the initial terms in (3). [it is easily checked that the determinant of
the k x k matrix (/3.) does not vanish, so that determination of the a.f s is
1970]
485
unique. ] Since the initial terms were not all zero, at least one of the a. f s
is non-zero.
Let p be the largest value of j for which a. f 0,
so that
p > 1. Then
Z-#
j*]
and
P"1 a.
V
1 -
a B
P P
p-i
p p
a.
1L
But
<1
for j = 1 9 2, , p - l j
and hence 9
lim .
n -*oo|
1 ,
pp
or equivalently,
lim f l n V - lnU j8 | n | =
n -*oo|^
n
I P P| J
(5)
Since p
is algebraic, it is easily verified that \p \ is also algebraic.
Moreover, |jS | f 1 by hypothesis so that ln|/3 | is irrational by application

of Lemma 2. But the sequence {n^}
is uniformly distributed mod 1 when-
ever f is irrational; therefore 9 the sequence
486
Dec. 1970
{nln|/ p |}^{ln|p p H
is uniformly distributed mod 1 and the same is true for the sequence
MPiK,r>rFrom (5) and Lemma 1, it is then clear that (in V } is uniformly distributed
" i
mod 1 as asserted, q. e. d.
The specialization to sequences satisfying the Fibonacci recurrence,
U
= U
- +U
(n > 1) tJ is immediate since the relevant polynomial in
this case is x - x - 1, and there are two distinct roots of unequal magnitude, namely
1 %/5
2
From the theorem, we conclude (in U }
is uniformly distributed mod 1
independently of the (non-zero) values specified for Uj and U2.

Lastly, we give an example to show that our assumption on the roots of
the associated polynomial cannot be relaxed. Consider the recurrence V
= V
for n > 1 with V4 == 1, V2 = 1. Then clearly V
= 1 for all n > 1
so that (in V } is a sequence of zeroes and hence not uniformly distributed

mod 1. The associated polynomial in this case is x2 - 1 which has two distinct roots, 1; however, the roots have magnitude unity, and therefore, the
conditions of our theorem are not met.
REFERENCES
1. I. Niven, 'Irrational Numbers," Carus Mathematical Monograph Number
II, The Math. Assn. of America, John Wiley & Sons, Inc. , N. Y. , 1956,
2. L. Kuipers, "Remark on a PaperbyR. L. Duncan Concerning the Uniform
Distribution mod 1 of the Sequence of Logarithms of the Fibonacci Numb e r s / ' TheJ[iboji^^
Vol. 7, No. 5, Dec. 1969, pp. 465-466,
473.
3. P. R. Halmos, Measure Theory, D. Van Nostrand Co. , Inc. , N.Y. , 1950.
ADVANCED PROBLEMS AND SOLUTIONS

Edited by
RAYMOND E.WHITNEY
Lock Haven State College, Lock Haven, Pennsylvania
Send all communications concerning Advanced Problems and Solutions
to Raymond E. Whitney, Mathematics Department, Lock Haven State College,
Lock Haven, Pennsylvania 17745,,
This department especially welcomes
problems believed to be new or extending old results. Proposers should submit solutions or other information that will assist the editor.
To facilitate
their consideration, solutions should be submitted on separate signed sheets

within two months after publication of the problems.
H-l 75
Proposed by L. Carlitz, Duke University, Durham, North Carolina.
Put
( l + z + i z2\
-n-1
n
a(n,k)zk
=J]
k=0
Show that
/tv
(!)
,
v
2*5*8 . . . (3n - 1)
a(n,n) =
,
S
258 . . . (3n - 1)
nl
(;)(")(-')'
s=0
^ E (n r r )( 2 V ") ^
^^
r=0
where
co = | (-1 + N P 3 ) .
487
2*58 - - (3n - 1)
n!
488
H-l 76
[Dec
Proposed by C. C. Yalavigi, Government College, Mercara, India.
In the "Collected Papers of Srinivas Ramanujan," edited by G. H.

Hardy, P. V. Sheshu Aiyer, and B. M. Wilson, Cambridge University P r e s s ,
1927, on p. 326, Q. 427 reads as follows:
Show that
Jm
e
-1
4ir
e
6TT
-1
24
Sir
-1
Provide a proof.
H~l 77
Let R(N) denote the number of solutions of

N
+ F
ki
k2
'
+ F
(r
kr
lj
'
' *") '
where
kj > k2 > > k r > 1 .
Show that
(1)
R(F
2nF2m>
2n+lF2m) =
(n
(2)
R(F
2nF2m+l) =
(3)
R(F
2n+lF2m+l) =
(4)
R(F|n+1) = R(F|n) = F ^
"
(n
m)F
(n
"
"
m)F
2m
+ F
2m-1
2m+l
m)F
2m+l
SOLUTIONS
SUM INVERSION
H-151
a.
Put
(n
m)
(n > m) ,
<n >
m)
(n>l)
'
.
1970]
489
00
2 , - 11
(1 - ax - bxy - c y ) "
V
A
x m y11
J
JLd
m,n
m,n=0
Show that
00
_j
X n9n
xR =
n=0
B.
t "
2bx + (fc>2
4ac x2
> >
Put
(1 - ax - bxy - cy)""1 =
J^
m 9 n=0
V n *m n
y
Show that
j B
x
n,n
n n
n=0
= {(1 - bx)2 - 4acx}~ 2
Solution by M. L. J. Hautus and D. A. Khmer, Technological University, Eindhouen, the Netherlands.
In a paper submitted to the Duke Mathematical Journal (the diagonal of

a double power series) 9 we have proved the following result:
Theorem.
Suppose
F(x,y) = J ]
^f(m,n)xmyn
m=0 n=0
converges for all x and y such that | x | < A, |y| < B, then for all z such
that |z| < AB, we have
490
[Dec.
00
m /F(s,z/ S )f = 2 f ( n > n ) z n '
n=Q
where C is the circle { s : | s | = (A + | z | / B ) / 2 } . Furthermore, if F ( s , z / s ) / s

has only isolated singularities inside C, then the integral can be evaluated
by summing the residues of F ( s , z / s J / s at these singularities. Coincidentally, we gave Carlitz 1 Problem B as an example in our paper,, Problem A can
be treated in just the same way. According to the theorem cited,
2n = 0 An,nx11 " SrT /C ;as 4
-s ds
- (1 - bx)s 2 + ex2
The singularities of the integrand are

I
1 ._ bx - (1 - 2bx + b2x2 - 4acx 2 )M
71 _ |
(
__
_J
and
,u
, (
bx + (1 - 2bx + b ^ 2 - 4acx 2 ) 2 V
and the singularities 6^ tend to 0 with x while the singularities 0 2 do

not.
Thus, the contour C must include 8f and exclude 6%; using the
residue theorem, we easily calculate

1
2771
2iri
-sds
IJ a(s - ei)(s.+,ei)(s - 0-2)(s + e2)
Substituting the values of ^
a(0|
_ 02}
and 6 2 given above yields the desired result.
A generalization of Problems A and B can be given as follows:
1970]
491
Let
F(x,y) = ] P f ( m f n ) x m y n
m,n
= (1 - ax k - bxy - cy k )
and let
f(n,n)x n
F(x) = J^
Then according to the theorem cited above, we have

k-1 ,
F(x)
Set co = e
j = 1, . . .
27ri
lift pJJ
as
2k
, , k ^
k
- M(1 - bx}$ + ex
' , then the singularities of the integrand are ^0l9^62.

9
for
k, where
kih
71
1/k
1 - bx - (1 - 2bx + b2x2 - 4acx )
" 1
25
"*
~
2 2
k 2
1_ - bx + (1 ,- 2bx
_ + b_ x - 4acx )
Since 0j_ tends to 0 with x and 62 does not, C includes the singularities
u^fij for j = 1,
8ee
, k, but excludes the singularities fcp02 for j = l 9 - # ,
k. Now we have the residue theorem to find that
i?( ) = -A277i
i
~s
ds
I 7k
^ T 7 ^ 7&7
(s - 0 4 )(s - 6 2 )a
(1 - 2bx + b2x2 - 4acx )
1
2trik
f
;
k
/ ,,
kk
J (t - flGj) ( t - 0 ) a
a(0^-0k)
492
[Dec.
k
where C? i s a contour In the t-plane which encircles the singularity 0 j , k
k
times but excludes 0 2
Also solved by D. V. Jaiswal and the Proposer.
HIDDEN IDENTITY
H-153 Proposed by J. Ramanna, Government College, Mercara, India.
Show that
G>
<>
16
F3k+1F3k+2(2F2k+1
F 6 k + 3 ) ( 2 F 3 k + 2 +. F ^ )
3 k + l F 3 k + 2 F 6 k + 3 ( 2 F 6 k + 3 " F 3k F 3k + 3> "
F ^
* 3n+3
2r
Hence, generalize (i) and (ii) for F + o Solution by the Proposer.
We note that (i) and (ii) a r e easily verified for k = 0 and k = 1 and
assume the results for k = r and prove them for k = r + 1. Thus we need
show, on subtracting (i) and (ii) for n = r from (i) and (ii) for n = r + 1,
respectively, that
(i)
4F
3(r+l)+l F 3(r+l)+2
= F6
(2)
(ii)
16
(2F
- F6
3(rH)+l
+ F
6(r+l)+3 ) ( 2 F 3(r+l)+2 + F 6(r+l)+3 )
*3(r+l)+3
*3r+3
F 3 ( r + 1 K 1 F 3 ( r + 1 ) + 2 F 6 ( r + 1 ) + 3 ( 2 F | ( r + 1 ) + 3 - F23(r+1)F23(r+1)+3)
- F8
- F&
~ *3(r+l)+3
*3r+3
Equations (1) are true since

(i)
(3)
4F
3(r+l)+l F 3(r+l)+2 (2F 3(r+l)+l

=
(F
3(r+l)+2
+ F
+ F
6(r+l)+3 )(2F 3(r+l)+2 + F 6(r+l)+3 )
3(r+1)+1 )6 "
(F
3(r+l)+2 " F 3(r+1)+1 )
1970]
(ii)

16 F 3 ( r + 1 ) + 1 F 3 ( r + 1 ) + 2 F 6 ( r + 1 ) + 3 ( F 4 3 ( r + 1 ) + 2 +
6F
+ F
= F
3(r+l)+3 " F 3r+3
16F
493
3(r+1)+2F3(r+lHl
3(r+1)+1 )
3(r+l)+l F 3(r+l)+2 F 6(r+l)+3 ( 2 F 6(r+l)+3

Jj
3(r+l) J J 3(r+l)+3 ;
Hence, the desired results follow.

Also solved by D. V. Jaiswal and C. C. Yalavigi.
TRIPLE THREAT
H-154
Show that for m 5 n , p integers >0,
/ m + l \ / n + l \ / p + l \
\i + k + l / \ i + k + l / \ i + 3 + 1 /
E
i,j,k>0
m
= EEE( m -r b )( a "c + c )( p -a + a )
a*t> b=0 c=0
and generalize.
Solution by the Proposer.
Put
S
m
l m + 1 \ / n + l \/ p + 1 \
\ j + k + \)\i + k + l / \ i + 3 + 1 /
'
i,j,k>0
Tm,n9p
Then
m
n
p
B \ T ^ V ^ \ ^ / n i - a + b \ f n - cb + c V p - c + a\
L L H
a=0 b=0 c=0
494
[Dec.
00
xmynZP
d>i
m,n,p=0
m,n,p
X^
j+k i+k
i+j x ^
/m + j + k + l\
L h
i,j,k=0
k+1 j
m=0
m Y"* / n + i
k + 1\ n
L, \ i + k 4-1 ft
n=0
00
#
y^ /p + i + j + i\ P
z
Z^r l
i+ j + i I
p=0
00
i+k i+k i+jJ / n

-j-k-2,v-i-k-2,v - i -J i - 2
xJ
y
z ( l - x) J
(1 - y )
(1 - z)
i,j,k=0
(i - x)-2d - x)-2d - Z )- 2 (i. T T _ j | _ ) y 1 ( 1 - T r r | ^ ) " 1

(x - TrrxXHi-x))
= (1 - y - z) V
- x - z)
( l - x - y) * .
In the next p l a c e ,
00
m,n,p=0 T
m
m,n,p
p
zr
a,b,c=0
m=0
n=0
p=0
CO
c,v-b-1/v-c-1/vz (1 - x)
(1 - y)
(1 - z)
a,b,c=0
- a - x,-V- -ia -_,,-(, - ^

= (1 - x - z ) "(1 - x - y )
(l - y - z )
(x - ^ " ' ( i -
1970]
495
and the r e s u l t follows at once*
GENERALIZED VERSION
Let k > 2
and n l 5 n 2 s
1 +
^\\+
ees
n2
i + vW
(^
n k non-negative i n t e g e r s .
+a
2+ V
+ a A /a3 + a4\
"M
nk
\\-i+
/ a ^
3 /
Show that
+ a2]\
2k-l
/
e
w h e r e the f i r s t summation i s o v e r all non-negative a l 9
, a k while in the
second s u m
2k
+ a
= n
l'
Solution.
+ a
3 = V
+ a
L e t S(n l 9 * , 1%)
denote the second sum 6
= n
3 ' " ' a2k-2
+ a
2k-l
denote the f i r s t s u m and T ( n l s , % )
C o n s i d e r the expansion of
-1
</) = </)(xl5 , x k ) = (1 - xj - x 2 )
-1
(1 - x 2 - x 3 )
-1
(1 - x k - x 1 )
Since
(1 _ x - y)"1 = ((1 - x)(l - y) - xy)"1 =

a=0
- _x
(1
x)
a
(1
-^
-
y)
we have
x>X!^...>1+k
a ^ - . a ^ O
(1 -
X l
a +a +2
a +a +2
2
) k 1 (1-x,)1
. . . ( l - x
\-l+ak+2
) k l
496
E
l'""'
,a
i+a0
a.
rV
=0
V^
b
i,+a1
*2
(\
a
^
V
l''"'bk=0
'
n +a.
\/ a l
+ a
"'A v i
^,,^=0
n..l
X
n2
X
/ V
/Vi
+ a
+ a
+ b
+ x
k + bk + 1 \
+a
+1
\
/
i ,b2
x
/ !
Tc
X
S(n
2 ' k
l ' V *'
On the o t h e r hand, s i n c e
a,b=0
ra2=V
'
3'a4=V
E
n
l,'"'nk=:()
n .l
xx
+ a
T(ni,n2,...,nk)
2k-l,a2k~
n2
x2
'
/ a 2k-l
V^
a
k
...x
The s t a t e d r e s u l t now follows at once.
2k\
2k-l
1970]
497
RECURRING THEME
H-155
Proposed by M. N. S. Swamy, Nova Scotia Technical College, Halifax, Canada.
The Fibonacci polynomials are defined by
W X ) = rfnW
n-l ( x )
with fi(x) = 1 and f2(x) = x. Let z

= f (x)f (y). If z
satisfies the
TjS
r s
r, s
relation
z
. + a z + bz n n + cz ^ - + dz
= 0
r+4 5 s+4
r+3 3 s+3
r+2,s+2
r+l,s+l
r,s
show that
b = -(x 2 + y2 + 2)
a = c = -xy,
and d = 1
Let u r = f r (x) and v r = f r (y).

Z
r+4,s+4
V4V4
= (xu
yzr+3,s+3
r+3
+ Z
Then,
+
V2
r+2,s+2
) ( y v
+ (xu
s+3
+ V
r+3vs+2
s+2>
+ y
V2Vs+3)
Now,
(xu
r+3VS+2+yUr+2vs+3)
= x(xur+2 + u r + 1 ) v s + 2 + y(yvs+2 + vg+1)ur+2
= (x2 + 3 ^ 2 , 8 + 2
+ (x
VlV2
yVlV2
= (X2 + y ) z r + 2 j S + 2 + x U r + 1 ( y v s + 1 + v g ) + y v s + 1 u r + 2
= (x2 + y 2 ) V 2 , s + 2
^Vl.s+l
+ xu
= (x2 + y 2 ) ^ r + 2 ) S + 2
^VllS+l
r+lvs
V2(vs+2 " Vs)
^ ^ V l " V2 )
= (X2 + y2 + l ) Z r + 2 } S + 2 + x y z r + l i g + 1 + v g (-u r ) =
+ z
r+2)S+2
498
= fr* + y* + D z r + 2 ) S + 2 + x y z
r + l j S + 1
+ zr>g
Dec. 1970
Hence,
z
,/i
, / i
r+49s+4
= x
V
+ (x2 + Jy 2 + 2)z l 0
+ x yJ z ,,
J Z l 0
r + 3 , s +l 30
r + 2 , s +l 20
r + l , s +nl
- z
r,s
Thus,
a = -xy,
b = -(x 2 + y2 + 2),
c = -xy,
d = 1 .
y4fco solved by W. Brady, D. Zeitlin, andD. V. Jaiswal.
L a t e Acknowledgement: D. V. J a i s w a l solved H - 1 2 6 , H - 1 2 7 , H - 1 2 9 , H-131.
LETTER TO THE EDITOR

DAVID G. BEVERAGE
San Diego State College, San Diego, California
In r e g a r d to the two a r t i c l e s ,
A S h o r t e r P r o o f , " by Irving Adler (De-
c e m b e r , 1969, Fibonacci Q u a r t e r l y ) , and
1967as the Sum of T h r e e S q u a r e s , "
by B r o t h e r Alfred B r o u s s e a u (April, 1967, Fibonacci Q u a r t e r l y ) , the g e n e r a l

r e s u l t i s as follows:
x 2 + y 2 + z 2 = n i s solvable if and only if n i s not of the form 4 (8k +
7),
for t = 0, 1, 2, , k = 0, 1, 2 , *
Since 1967 = 8(245) + 7,
1967 f- x 2 + y 2 + z2
to F e r m a t and proven by D e s c a r t e s i s that no i n t e g e r
A l e s s e r r e s u l t known
8k + 7 i s the s u m of
t h r e e rational squares,,* *
* William H0 Leveque* T o p i c s in N u m b e r T h e o r y , Vol. 1, p. 133 e
* * L e o n a r d E . Dickson, H i s t o r y of the T h e o r y of N u m b e r s , Vol. II, Chap,
VII, p. 259.
DETERMINATION OF HERONIAN TRIANGLES

JOHN R.CARLSON
San Diego State College, San Diego, California
1.
A P y t h a g o r e a n t r i a n g l e i s defined a s any r i g h t - t r i a n g l e having i n t e -
gral sides.
Using the well-known r e l a t i o n s h i p
a2 + b2 = c 2
where
a,
b,
and c a r e the two s i d e s and the hypotenuse r e s p e c t i v e l y , it i s obvious t h a t

one of the s i d e s m u s t be even.
integer,,
H e n c e , the a r e a of such a t r i a n g l e i s also an
In h i s book* , O r e i n t r o d u c e s the g e n e r a l i z a t i o n of this situation: a
t r i a n g l e i s called Heronian if it h a s i n t e g r a l s i d e s and a r e a . He further c o m m e n t s that, "although we know a c o n s i d e r a b l e n u m b e r of Heronian t r i a n g l e s ,

we have no g e n e r a l formula giving them all. ? ? In t h i s p a p e r , we p r o p o s e to
find all such t r i a n g l e s and p r o v e a few b a s i c p r o p e r t i e s concerning t h e m .
2.
Since e v e r y P y t h a g o r e a n t r i a n g l e i s H e r o n i a n , and since
Pytha-
g o r e a n t r i a n g l e s a r e completely d e s c r i b e d by the well-known f o r m u l a s for

the s i d e s u 2 + v 2 , u 2 - v 2 ,
and 2uv, the r e a l p r o b l e m i s to c h a r a c t e r i z e all
n o n - r i g h t - a n g l e d Heronian t r i a n g l e s .
L e m m a 1.
Let
a, b , c ,
with s i d e s of n a , n b ,
(1)
and n all be i n t e g e r s .
Then the t r i a n g l e
and nc i s Heronian if and only if the " r e d u c e d " t r i -
angle with s i d e s a , b , c
Proof.
We f i r s t give an obvious p r o p e r t y .
i s Heronian.
We shall u s e Heron 1 s f o r m u l a for the a r e a of a t r i a n g l e

A = -JsTs - a)(s - b)(s - c ) ,
where
s = - (a + b + c) .
L e t A be the a r e a of t r i a n g l e a , b , c
and l e t Af be the a r e a of
n a , n b , nc e
Then Eq. (1) shows u s i m m e d i a t e l y that
*Oystein O r e , Invitation to N u m b e r T h e o r y , pp. 5 9 - 6 0 , Random H o u s e , New

York, 1967.
499
500
[Dec.
Af = n2A .
(2)
Hence, if A is an integer, so is AT.
For the converse, suppose that Af
is integral. Then Eq. (2) insures that A is at least rationaL
On the other
hand, Eq. (1) implies that A is the square root of an integer, which is well
known to be either integral, or irrational.
Thus we conclude that A must be
an integer and the lemma is proven.

Before we proceed to our first theorem, we illustrate with two examples.
Suppose we juxtapose (or "adjoin") the two Pythagorean triangles 5, 12, 13
and 9, 12, 15 so that their common-length sides coincide. Clearly a (nonright-angled) Heronian triangle results with sides of 13, 14, 15 and area
equalling 84.
As a second example, we adjoin the triangles 12, 16, 20 and 16, 63, 65
(one of which is primative) to obtain the Heronian triangle 20, 65, 75 which
may be reduced to 4, 13, 15, a primitive Heronian triangle with area equalling 24. That these two examples illustrate all possible events is the content
of our first theorem.
Theorem 1. A triangle is Heronian if and only if it is the adjunction of
two Pythagorean triangles along a common side, or a reduction of such an
adjunction.
Proof. One direction is clear:
since every Pythagorean triangle is
Heronian, so is every adjunction of two.
The previous lemma guarantees
that every reduction is also Heronian,

Thus, let us suppose that the triangle a , b , c
is Heronian and try to
prove that it is either the adjunction or a reduction of an adjunction of two

Pythagorean triangles. First we assume the obvious: that from some vertex
we may draw a perpendicular to the opposite side, thus dividing the given t r i angle into two "adjoined" right triangles.
That such is possible is easily
shown. Let the length of the altitude so constructed be x and let the base c
be thus divided into segments c1 and c 2 ,
triangle is Heronian, it is clear that
is rational.
Let
so that c = cA + c 2 .
Since the
1970]
501

x
m
=
n
be reduced to lowest terms.

By the law of cosines* b2 = a 2 '+ c2 - 2ac cos wi and thus
C0S w
is rational.
a2 + c2 - b2
i =
Wc
Hence both Cj = b cos wj and c2 = c - cA are also rational*
If the numbers x, c 1? and c2 are in fact all integers, we are done. Otherwise we look at the triangle having sides na, nb, and nc. From elementary
geometry, this triangle is similar to the original one and thus the new altitude
is equal to nx = m ,
an integer,, But nc 1? still rational, is given by
v^
which is 9 as before, either integral or irrational, and thus must be integraL
Likewise nc 2 is integrals and the new enlarged triangle is the adjunction of
two Pythagorean triangles. Thus the original is a reduction of an adjunction
and we are through.
Since the sides of any Pythagorean triangle are given by u2 + v 2 , u2 v 2 , and 2uvs we now have a method for finding all Heronian triangles.
Corollary 1. A triangle is Heronian if and only if its sides are given
by either (3) u2 + v 2 , r 2 + s 2 ,
2
and u2 - v2 + r 2 - s2; where

2
r s = uv;
or
(4) u + v , r + s , and 2(uv + rs); where r - s = u - v ; or (5) a r e duction by any constant factor in either case (3) or (4).
3. Although the preceding theorem and its corollary give formulations
for finding all Heronian triangles, there are many properties of Heronian t r i angles that are not obvious from examination of the special subset of rightangled triangles. Some of these properties will be given here.
Lemma 2. A primitive Heronian triangle is isosceles if and only if it
has sides given by (3), (4), or (5) with r = u and s = v.
Proof.
Since a triangle is primitive only when one side is even, the
equal sides of the isosceles triangle must be odd? say 2m + 1. Let the even
502
DETERMINATION O F HERONIAN TRIANGLES
side be 2n.
Then the s e m i p e r i m e t e r i s given by
[Dec.
2m + n + 1 and the a r e a ,
given by Eq e (1) b e c o m e s
^f(2m"+ n + l ) ^ T T ^ T l J ( n j f o )
so that
A = n ) / ( 2 m + I) 2 - n 2 .
If t h i s i s to be an i n t e g e r , t h e r e m u s t be an i n t e g e r Q such that
(2m + I) 2 - n 2 = Q 2 .
T h u s the n u m b e r 2m + 1 i s the hypotenuse of a P y t h a g o r e a n t r i a n g l e , which
m e a n s j of c o u r s e , that
2m + 1 i s a s given in C o r o l l a r y 1.
Conversely,
e v e r y t r i a n g l e d e s c r i b e d by those formulae will be i s o s c e l e s if
r = u
and
s = v.
We note in p a r t i c u l a r t h a t any n u m b e r of the form
4n + 2 m a y be used
a s the even side of a p r i m i t i v e i s o s c e l e s Heronian t r i a n g l e , by u s i n g s i d e s

2n 2 + 2n + 1,
2n 2 + 2n + 1,
and 4n + 2, We will show* in fact, that any i n t e -
g e r g r e a t e r than two m a y be u s e d a s the side of a p r i m i t i v e n o n - r i g h t - a n g l e d

Heronian t r i a n g l e ,
Before we d o , we shall e s t a b l i s h the following*
L e m m a 3 9 No Heronian triangle h a s a side of e i t h e r 1 o r 20

Proof,
Since the s i d e s of the t r i a n g l e m u s t be i n t e g e r s , the difference
between two s i d e s i s e i t h e r 0 o r an i n t e g e r ^>1, This l a t t e r c a s e would p r e clude the u s e of 1 a s a s i d e .
But an i s o s c e l e s t r i a n g l e with side one i s also
i m p o s s i b l e j since in that event, we m u s t have two s i d e s equal to 1, and hence

the t h i r d side e i t h e r 0 o r 2* Thus we have only to show that 2 cannot be used
a s the side of an Heronian t r i a n g l e . Suppose, to the c o n t r a r y , that we do have
a triangLe with s i d e s 2, a,
and b,
Then the a r e a a s given by (1) i s
A - \ s ( s - 2)(s - " a M s ^ T ) ,
where a + b + 2 = 2s,
The only v a l u e s of a and b which satisfy t h i s l a s t
equation a r e a = b = s - 1.
T h u s the a r e a b e c o m e s
1970]
503
A = Vs(s - 2)(1)(1)
and we m u s t have s(s - 2) = Q 2 for s o m e Q* But t h i s Is impossible* so we
a r e done e
Using the formulae for the s i d e s of a P y t h a g o r e a n t r i a n g l e , It Is e a s y
to show that e v e r y i n t e g e r g r e a t e r than two can be u s e d a s a side In a finite
n u m b e r of P y t h a g o r e a n t r i a n g l e s .
This o b s e r v a t i o n h a s the following r e -
m a r k a b l e generalization,,
T h e o r e m 2a
L e t a be an i n t e g e r g r e a t e r than two e
Then t h e r e e x i s t s
an infinitude of p r i m i t i v e Heronian t r i a n g l e s with one side of length a e

Proof,
If a i s odd* we m a y u s e s i d e s given by
(6)
a 9 i ( a t - 1),
and
| ( a t + 1)
w h e r e t i s a solution of the P e l l i a n equation

(7)
t 2 - (a2 - l)y 2 = 1 .
Since a 2 - 1 i s even and n e v e r a perfect s q u a r e , Eq* (7) h a s an infinitude of

solutions for t ,
each of them odd 9 so that (6) l i s t s only Integers,,
Since
| ( a t - 1) + 1 = | ( a t + 1),
the t r i a n g l e i s obviously primitive*
We compute the a r e a of the t r i a n g l e by
(1) and find

A 2 = | ( t + l ) a | ( t - l ) a | ( a + l ) | ( a - 1) = (f) 2 a 2 (a 2 - l)(t 2 - 1) .
But, by Eq* (7), we have t 2 - 1 = (a2 - l)y 2 so that
A2 = (|) 2 a 2 (a 2 - l) 2 y 2
and thus
A = -g-(a - D ^ ( a + D a y ,
504
[Dec.
an integer, and hence the triangle is Heronian.

Next, suppose that a is even, say a = 2n,
where n is odd.
Then
we may use
(8)
a, tn - 2,
and
tn + 2 ,
where t is any odd solution of

t2 - (n2 - 4)y2 = 1 .
(9)
Since n is odd, n2 - 4 is also odd and thus an infinitude of odd values of t

is available. That (8) forms a primitive triangle is clear; we prove it is
Heronian by computing
A2 = (t + l)n(t - l)n(n + 2)(n - 2) = n2(n2 - 4)(t2 - 1) = n2(n2 - 4)2(y)2
so that A is an integer,
Lastly, suppose that a = 2n, where n is even. Then we may use
(10)
a, tn - 1,
and
tn + 1,
where t is any solution of

t2 - (n2 - l)y 2 = 1 .
(11)
The proof follows the same lines as that just given.
Thus our theorem is
proven for all cases.

It should be obvious that the formulations given in the above proof are
simply chosen from an infinitude of possibilities.
shown, for example, that the triangle having sides
a, ^-a(x - 1) + 1,
and
Thus we could also have
1970]
505
~ a ( x + 1) - 1,
w h e r e x i s found from
x 2 - (a - l)y 2 - 1 ,
which will have an infinitude of solutions a s long a s
a - 1 i s not a p e r f e c t
square,
4.
We conclude with a few o b s e r v a t i o n s and a s h o r t l i s t of examples.,
It i s e a s y to show that e v e r y p r i m i t i v e H e r o n i a n t r i a n g l e h a s exactly one even

side.
A s i m p l e check of all the p o s s i b i l i t i e s obtained from the adjunction of
two P y t h a g o r e a n t r i a n g l e s (which a r e known to have e i t h e r one o r t h r e e even

sides) will suffice to p r o v e t h i s . F r o m t h i s , we conclude t h a t the a r e a of any
Heronian. t r i a n g l e i s divisible by 2, s i n c e , if s i s e v e n , a factor of 2 divides
A 2 (and thus also A) w h i l e , if s i s odd, then s - a will b e even w h e r e
i s an odd side of t h e t r i a n g l e , and so again 2 divides the a r e a ,

Since the a r e a of any P y t h a g o r e a n t r i a n g l e i s given by
A = uv(u - v)(u + v)
i t i s e a s y to show that such a t r i a n g l e h a s a r e a divisible by t h r e e ,
A simple
a n a l y s i s of adjunctions and p o s s i b l e r e d u c t i o n s will then show that
every
Heronian t r i a n g l e h a s a r e a divisible by t h r e e ,
Following i s a l i s t of the " f i r s t " few p r i m i t i v e (non-right-angled) H e r o n ian t r i a n g l e s :
a = 3
b = 25
c = 26
13
15
29
30
25
29
7.
15
20
29
35
506
_10
= 17
65
70
10
13
13
10
17
21
11
13
20
12
17
25
13
13
24
13
14
15
13
20
21
13
37
40
14
25
25
15
28
41
15
37
44
15
41
52
16
17
17
17
17
30
17
25
28
18
41
41
19
20
37
20
37
51
21
85
104
22
61
61
23
212
225
24
37
37
25
25
48
25
29
36
25
39
40
25
51
74
26
85
85
27
676
701
28
85
111
29
29
40
29
29
42
17
[Continued on page 5510 ]
[Dec.
PINEAPPLES MB FIBONACCI NUMBERS

PHILIP B.ONDERDONK
Philadelphia, Pennsylvania 19118
In the book? Fibonacci and Lucas Numbers, by V. E. Hoggatt9 J r .

there is a paragraph mentioning the spiral curves on pineapples which makes
reference to Mathematical Diversions by Hunter and Madachy (Princeton.s
No J. , published by D. Van Nostrand Co5 Inc., in 1963). Thus* parastichies
on pineapples were known9 then 9 sometime prior to 19638
The author first became aware of the spiral curves and the agreement
with Fibonacci numbers about 1951 upon reading an article in a magazine put
out by the American Association for the Advancement of Science which pertained to Fibonacci Numbers. At that time* the author was working for the
Maiu Pineapple Company, Kahului ? Maui ? Hawaii, and after reading this
article s started checking pineapples. The vast majority of pineapples checked
had 8 - 1 3 - 2 1 rows of fruitlets (eyes) A few runts were
5-8-13.
Since a pineapple with more fruitlets for a given size would likely have
a finer texture and would be better for eatings the author was interested in
finding a pineapple with 13 - 21 - 34 rows.
No such pineapples were ever
found5 however.
Giving evidence to the fact that numbers of spiral rows on pineapples
were studied even earlier, the Experiment Station of the Association of
Hawaiian Pineapple Canners published an article on such a study as early as
1933e The article by M. B. Linford was published in the Pineapple Quarterly,
Vol. HI, No. 49 December 1933, pp. 185-195. It was entitled "Fruit Quality
Studies n , Eye Number and Eye Weighty and mentioned the number of rows.
of fruitlets on pineapples as basically 8 - 1 3 spirals, ranging from 5 - 8 - 1 3
- 21 for the several types of spirals. The article did not s however, draw any
connection between these numbers and the Fibonacci Numbers.
Following is a quotation from a section of this article which had the
headings
A Method of Estimating Eye Numbers. "

"A Cayenne pineapple fruit of usable size consists of from 100 to
200 fruitlets. This makes counting eyes in any large numbers of fruits
both tedious and expensive. Where great precision is required there is
no short cut* and when counts are made of very young inflorescences
long before flowering the use of a low-power microscope is essentiaL
507
508
PINEAPPLES AND FIBONACCI NUMBERS
Dec. 1970
"For use where such precision is not required, however, the

regular arrangement of eyes (fruitlets) suggested a method of estimating which is much more rapid. As may be seen in Figure 1, the fruitlets are aligned in two series of spirals, one rising to the right, the
other to the left. One series of spirals is steep and is found, in Cayenne
and at least some other varieties, to be composed of 13 separate rows
of eyes. The other s e r i e s , less steep, is composed of 8 longer spirals.
Various irregularities occur, but in examining many individual fruits
over a period of two years, no specimen has yet been found in which
the basic pattern was other than 8 and 13 spirals. It follows from this
uniformity that a count of eyes in one spiral, multiplied by the number
of similar spirals should yield a figure close to the actual number of
eyes. E r r o r s are introduced by two factors: Some spirals have more
eyes than others of the same fruit. In Figure l b some spirals contain
11 eyes, some contain 12. Chance will determine which spiral is
counted. Then as shown in Figure 2, irregularities may result in there
being more or less than the regular number of spirals through part of
the length of the fruit. By actual test, it has been determined that
smaller e r r o r s result from counting the longer spiral and multiplying
by 8, than from counting the short spiral and multiplying by 13. For
this reason, the standard procedure adopted is to count the long spiral/
On some fruits, this ascends to the right of the observer, on some to
the left. After a little practice, it is recognized readily as the less
steep of the two spirals which bound the four sides of an eye, considering the eye as a square standing on one corner. In case of doubt, it is
safest to verify the number of spirals before counting eyes.
"Confusion may arise with extraordinarily small or large fruits
from the fact that two other series of spirals may sometimes be recognized. These include steep, nearly vertical spirals, of which there are
21, and very flat spirals of which there are only 5. Thus the numbers
of spirals of the several types are 5, 8, 13 and 21. The number of eyes
counted in any spiral must be multiplied by the number of similar
spirals.
"In other plant parts where spiral patterns occur, chance alone
usually determines whether the spiral winds to the right or to the left.
This was tested for pineapples by recording direction of spirals along
with eye number for a large number of fruits during seasin 1932. Results are shown in Table 1. If chance determines we should expect to
find equal numbers of right and left patterns. Actually among 9,008
fruits, right spirals were found 49.4 percent of the time, left spirals
50.6. Only one of the lots shown in this table deviated markedly from
the expectation, and the 24 separate plots in that group showed no agreement among themselves; some had more left spirals and some more
right. Chance alone seems to determine the direction of these spirals
in the pineapple,"
THE POWERS OF THREE

J. M. WILLIAMS, JR.
San Francisco, California
Any number may be expressed in powers of three by addition or subtraction of the numbers those powers represent.
13 = 3 2 + 3 1 + 3
14 = 3 3 - 32 - 3 1 - 3 .
The powers used in such expressions are whole integers* no fractional powers
being involved.
The number of terms required to express a number approximates twice
the number of digits in the number; the greater the number of digits required
the more closely this limit is approached.
Any such expression of a number need contain no repetition of any given
power.
Such expressions are easily handled in arithmetic processes by observation of algebraic rules regarding exponents.
Discussion of this digital system follows.
It will be noted from Table 1 that the powers of three follow a routine
sequence in the expressions for the numbers, appearing first behind the
positive sigh, then changing to the negative sign, and then disappearing from
the statement.
The appearance of powers in the statements follows a fixed sequence:
1. The power appears for the first time at a number equal to one-half of the
value of that power with 1 added.
2. It appears behind the positive sign.
3. It remains in the statement, and behind the positive sign, for a series of
statements equal in number to the value of that power of three.
4. It then becomes negative in the following statement.
5. It remains in the statement, and behind the negative sign, for a series of
statements equal in number to the value of that power of three.
6 It then disappears in the following statement.
509
510
THE POWERS OF
O F THREE
THREE
[Dec.
Table 1
N u m b e r e x p r e s s e d in
P o w e r s of T h r e e
+ 3
1
1
+ 3
+ 3*
- 3
+ 3 1 + 3
4
2
- 3 1 - 3
+ 3
+ 32 - 31
+ 3 2 - 3 1 + 3
+ 32
+ 3
10
+ 32
11
+ 3 2 + 3 1 - 3
12
+ 32 + 31
13
+ 3 2 + 3 1 + 3
14
+ 3 3 - 3 2 - 3 1 - 3
15
+ 33 - 32 - 31
16
+ 3 3 _ 3 2 - 3 1 + 3
17
+ 3 3 .. 3 2
+ 33 _ 3 2
18
- 30
+ 3
- 30
7e It remains out of the statement for a series of statements equal in number

to the value of that power of three*
8 It then reappears in the statement., and behind the positive sign s and r e peats the sequence outlined above without limit.
Determination of the proper statement for any given number is outlined below*
Determination of Statement
1. Subtract 1 from the given number*
2, Divide the remainder by 3 9 setting the quotient below the dividend and
the remainder from the division to the rights whether this remainder be
2, or 1, or 0S
1970]
3e
THE POWERS O F THREE
511
If the r e m a i n d e r after division is 1 o r 2 p r o c e e d a s d i r e c t e d below,

but w h e n e v e r the r e m a i n d e r after division i s 0 it i s n e c e s s a r y to s u b t r a c t 1 from the quotient before p r o c e e d i n g to t r e a t it a s a dividend,
as
below,,
4* Divide again by 3 a s d i r e c t e d in step 2, and continue t h i s p r o c e s s until
the dividend i s 0 with 0 r e m a i n d e r , watching throughout the p r o c e s s
outlined in Step 3 0
5e
The column of r e m a i n d e r s 2 ,
o r 1,
o r 0, which have been s e t to the
r i g h t i s now n u m b e r e d , beginning at the top with 0 and p r o c e e d i n g with

1 ? 2, 3 f e t c o s and ending with the h i g h e s t n u m b e r in the sequence opposite
the final
0 remainder,.
The n u m b e r s in t h i s sequence a r e the p o w e r s of
three.
60
Fixation of signs for the v a r i o u s p o w e r s , o r exclusion from the s t a t e m e n t ,

i s d e t e r m i n e d by the r e m a i n d e r s :
When the r e m a i n d e r i s 0 the sign i s positive,,
When the r e m a i n d e r i s 1 the sign i s negative*
When the r e m a i n d e r i s 2 the p o w e r i s excluded from the statement,,
D e m o n s t r a t i o n of t h i s p r o c e s s follows:
Given n u m b e r
Divide by 3
6056
- 1
6055
Power
2018 with r e m a i n d e r
1 0
672
"
"
2 1
224
- 1
"
"
0 2
74
"
1 3
24
"
2 4
"
,?
0 5
tt
1 6
"
"
2 7
"
"
0 8
223
8
- 1
7
[Dec.
THE POWERS OF THREE
512
Consider the s.equence thus, bottom to top:

Remainders:
Signs:
Powers:
+
6
Statement for the given number:
Proof of statement:
Positive powers:
6561 + 243 + 9
6813
Negative powers:
- 729 - 27
-757
Given number:
6056
For simplicity, in further discussion, the digit 3 will not be used in power
statements except when the exponent 3 is. required..
only the digits designating the powers.
Statements will use
Treated thus the statement for 6056
would be written thus: 6056 equivalent, +8 -6 +5 -3 +2 -0..

ADDITION OF STATEMENTS
In handling two or more such statements arithmetically, it is almost
inevitable that there will be duplication of one or more of the powers.
Con-
sider the ad'dition of the statements for the numbers two and three:
2 equivalent,
+1-0
3 equivalent, +1
Addition of +1 to the statement gives +1 +1 -0..
+2 -1
-9-
This is re-written thus:
The next higher power is given the sign of the duplicated power,
following it by the duplicated power with reversed sign;
obviously 3 + 3 =
9 - 3.
When a power is triplicated, the next higher power is given the sign of
the triplicated power, which is then dropped from the statement.,
addition of the following statements:
Consider
197
THE POWERS OF THREE
513
2 equivalent, +1 -0
!?
3
+1
?f
4
+1 +0
9
+2
Note that the unlike signs calcelled and removed 0 power from the statement for the sum8
When the number of repetitions exceeds three, they can be eliminated
step-wise by application of the processes outlined above,,
SUBTRACTION OF STATEMENTS
Subtraction is performed by changing the signs on all of the powers in
the statement being subtracted and then performing addition as aboveL
MULTIPLICATION OF STATEMENTS
In performing multiplications the digits representing the powers are
added irrespective of sign, and the signs follow algebraic rule:
Multiplication of like signs yields the positive sign.
Multiplication of unlike signs yields the negative sign,,
Consider multiplication of statements for 13 and 14:
14:
+3 -2 - 1 -0
13
+2 +1 +0
+3 -2 - 1 -0
+4 - 3 -2 - 1
+5 -4 - 3 -2
+5 -4 +3 -2 +1 -0
Consider the totalling of those products step-wise:

1. The -0 comes down unchanged.
2. The duplicated - 1 becomes -2 +1
3* The triplicated -2 becomes - 3 , with the -2 carried forward from
Step 2 brought down unchanged.
4, The duplicated - 3 has been triplicated by the -3 carried forward from
Step 3 so it becomes -4 and the +3 is brought down unchanged.,
514
THE POWERS OF THREE
[Dec.
5. The unlike signs for 4 cancel so the -4 carried forward from Step 4 is
brought down unchanged.
6* The +5 is brought down unchanged.
Proof of statements
Positive powers?
243 + 27 + 3 =
Negative powers?
-81
-9
-1
273
-91
13 x 14 =
182
DIVISION OF STATEMENTS
This process brings another rule into play.
Before stating the rule,
consider these two t e r m s , + 4 - 3 and +3, the statements for 54 and 27.
If these terms are added, the cancellation of unlike signs will leave +4, the
statement for 81. If they are subtracted* the sign representing 27 would be
changed. It would then be duplicated.
After treatment for duplication, it
would cancel the +4 and leave +3, the statement for 27.
The rule in dividing power statements once more involves like and unlike signs, but with an extra specification added? like or unlike signs of
adjacent powers as in the case of the statement for
54.
The rule: when
adjacent powers in the dividend have unlike signs, the two powers must be
considered as a duplication of the lower power, both with the sign of the higher
power. This is a reversal of the rule for the additive process and is not unreasonable since division is a subtractive process.
Demonstration of division will be the reverse of the example given for
multiplication.
The statement for
182 .is divided by the statement for
below. The notes referenced are found on the following page.
+2 +1 +0
Rewritten
Rewritten
+2 +1 +1 - 0
+5 - 4 +3 - 2 +1 - 0
+4 44 +3 - 2 +1 - 0
-4
-3 -2
+4 - 3 +2 +1 - 0
+3 +3 +2 +1 - 0
-3 _
-2 -1
-0
-3 -2 -1
2 -1 -0
+2 +1 +0
See Note 1
See
See
See
See
See
See
See
See
See
Note
Note
Note
Note
Note
Note
Note
Note
Note
2
3
4
5
6
7
8
9
10
197
THE POWERS OF THREE
515
Notes:
1. In this case, ail the powers are both adjacent and unlike. The rule will
be applied only to the leading pair,
2. Rewritten giving the duplicated lower power the sign of the higher power.
3. The +2 in the divisor, when subtracted from the +4 in the dividend,
yields +2 for the quotient When the sign is changed for subtraction,
cancellation will result.
4. The duplicated -2 became -3 +2 and the +2 came down. Unlike signs
for 3 cancelled and the -3 brought forward came down. The unaffected +4 came down, making this another case of unlike signs for adjacent powers.
5. Rewritten as in Note 2.
6. The +2 in the divisor, when subtracted from the +3 in the remainder,
yields +1 for the quotient. When the sign is changed for subtraction,
cancellation will result.
7. Unlike signs cancel and +3 comes down with - 0 .
8. The +2 in the divisor, when subtracted from the +3 in the remainder,
yields +1 for the quotient. When the sign is changed for subtraction,
cancellation will result. The duplication of +1 in the quotient can be
cared for after completion of the division.
9. Unlike signs cancel, and unaffected terms come down.
10. The +2 in the divisor now encounters -2 in the remainder. Disregarding signs, the 2 from 2 yields 0 for the quotient. Adding -0 to
+2 will yield -2; when the sign is changed for subtraction, cancellation
will result.
The quotient reads +2 +1 +1 =0. There is duplication of +1. This is
changed to +2 - 1 . This change duplicates +2.
This is changed to +3 - 2 .
Now the corrected statement reads +3 -2 - 1 - 0 , the power statement for 14.
This is as far as I have investigated this curiosity with any success.
Perhaps someone else can find a way to extract roots and raise to higher
powers without simple multiplication. Decimal fractions can, of course, be
handled by appropriate multiplication by powers of 10, as can uneven divisions that result in significant remainders.
I apologize for using the expression "negative powers" when values involved are not reciprocals.
explanation.
This misuse was a short-cut through verbose
SOME REMARKS ON THE ORDERING

OF GENERAL FIBONACCI SEQUENCES
JOE R. CRENSHAW
Midwestern University, Wichita Falls, Texas
In The Fibonacci Quarterly, Vol. 1, No. 4, December, 1963, Brother

U. Alfred suggested a method whereby the general Fibonacci sequences
could be ordered. The remarks of this paper are intended to supplement,
rather than supplant, those of Brother Alfred.
Another system of ordering
is proposed herein.
We shall obtain aganeral Fibonacci sequence by taking any two integers,
a and b, and employing the relationship a + b c.
integers for indexing, with a = T
T
+-,
and b = T
Utilizing the set of
- , and requiring that T
<
we may define the general sequence by using the recursive form:

T
+ T
n
n+1
T
n+2
To eliminate possible confusion, we adjust the indexing such that T 0 is the

smallest non-negative term of the sequence.
It has been shown [1], and is easily verified, that when T 0 is the
smallest non-negative term of any general Fibonacci sequence,
2T 0 < T i ;
2Tn+1 > Tn+2,
(n = 0, 1, 2, 3, ) .
Thus, while any two successive terms of a general Fibonacci sequence are
sufficient to define the entire sequence, we should like to employ T 0 and Tj.
Hence, we have a unique representation of each general Fibonacci sequence,
i. e. ,
( T 0 , T i ) = T 0 , T l f T 2 , T 3 , T 4 ,
Attention is called to the fact that we do not require any two successive terms
to be relatively prime and employ the single restriction that T 0 be the s m a l lest non-negative term of the sequence.
Since we now have a unique representation of each general Fibonacci
sequence, one of the next logical steps would be to devise a method of ordering
516
Dec. 1970

517
the sequences. Such a method should be easy to apply, virtually by visual

inspection. Given any representation of a general Fibonacci sequence, we
immediately have the first two t e r m s , the third term is a simple summation.,
and we can easily calculate the characteristic number, D, from the wellknown relation [2]
T 2 j - T j T o - T20 = D .
From these few properties we should like to accomplish the desired ordering
of all general Fibonacci Sequences.
Utilizing the property that each sequence has a unique characteristic
number, D,
Brother U. Alfred has suggested a method of arranging the
general Fibonacci sequences with respect to the value of D.
Where several
sequences have the same D, the size of T 0 becomes the second criterion.
With the restriction that successive terms of the sequence be relatively prime,
no doubt to eliminate multiples of a sequence, he suggests the following
convention:
(To.Tj)
D
1
5
11
19
29
31
41
55
(0,1)
(1,3)
(1,4),
(1,5),
(1,6),
(2,7),
(1,7),
(1,8),
(2,5)
(3,7)
(4,9)
(3,8)
(5,11)
(6,13)
Let S' , (n = 1, 2, 3, 4, ) , denote the n
sequence of the ordering,
and we have
S[ = ( 0 , 1 ) ,
SJ = (1,3),
SJ = (1,4),
SJ = ( 2 , 5 ) ,
SJ = (1,5),
S^ = (3,7),
S\ = (1,6),
SJ = (4,9),
SJ = (2,7),
S'u = (3,8), .
518
[Dec.
By dropping the restriction that any two successive terms be relatively

prime, but retaining the same principle, we must then modify the above
method as follows:
D
1
4
5
9
11
16
19
20
25
29
31
36
41
44
45
Hence, we now have, with SM (n

quence of the modified ordering
(To.
(0,1)
(0,2).
(1,3)
(0,3)
(1,4),
(0,4)
(1,5),
(2,6)
(0,5)
(1,6),
(2,7),
(0,6)
(1,7),
(2,8),
(3,9)
Tj)
(2,5)
(3,7)
(4,9)
(3,8)
(5,11)
(4,10)
1, 2, 3, 4, ) denoting the n
se-
Si' = (0,1),
Sjf = (0,2),
S^! = (1,3),
S'i = (0,3),
ST5 = (1,4),
S'6f = (2,5),
Si1 = (0,4),
Si1' = (1,5),
Sg = (3,7),
Si?0 = (2,6), .
Let us examine three of the sequences, observing the above systems of

ordering.
S\
= Si3 = (4,9)
= 4, 9, 13, 22, 35, 57, 92,
D = 29
Sii = Sfi7 = (1,7)
= 1, 7, 8, 15, 23, 38, 61,
D = 41
Sf12 = siT8 = (5,11) = 5, 11, 16, 27, 43, 69, 112, D = 41 .

Here note that there is no simple relationship between the k
term of Sj
fT
and s! (or S. and S") as compared to the relationship of i and j , thus
1970]
519
eliminating one of the most desirable results we should like to have from a
system of ordering* For example? Given S23 and S ^
S23 smaller than the k
is the k
term of
term of sj 5 ?
Furthermore9 additional investigation has not revealed any simple r e th

lation for comparing k
terms of individual sequences using the readily
available information we should like to use for an ordering a Thus it appears,
at least for the present, we shall have to be content with a method of arranging the general Fibonacci sequences so that they maybe designated, or counted , for example?
Siofo = ( ? . ? )
or
(237,475) = Sff .
Utilizing a table of D ! s up to a value of 1000, one finds, for D greater

than 5, at least two sequences associated with each D* Occasionally, but
without a recognizable pattern? four sequences are found to be associated
with a particular D Therefore, without extensive calculation or the aid of
a lengthy table, we would not know how many sequences precede a sequence
having a D in, for example, the l g 000 t 000
region,,
Hence, an ordering,
using D as an index, becomes unwieldy with larger values of D, the characteristic number of the sequence*
With the above limitation in mind, an alternate proposal for ordering is
presented herein* There is no noteworthy advantage claimed, other than the
convenience of obtaining S
or (T 0 ,Ti) where large indices and/or initial

th
terms are involved* Again, there is no simple relationship between the k
th
term of S. as compared to the k
term of S..
Let us arrange the unique representation of each general Fibonacci s e n
quence, (T 0 ,Ti), in an infinite matrix array in the following manner(0,1)

(1,3)
(2,5)
(0,2)
(1,4)
(2,6)
(0,3)
(1,5)
.(2,7)
(0,4)
(1,6)
(2,8)
(0,5)
(1,7)
(2,9)
(3,23 + 1)
(3,23+2)
(j,2].+ 3)
(3,23+4)
(3,23 + 5)
9,
(3 = 0, 1, 2, 3, 4, )
*See [5] and [6].
520
[Dec
Given the above display of general Fibonacci sequence representations, the

remaining problem is that of choosing the system of ordering to be employed.
Note that the representation is in the T 0 + 1 row and the TA - 2T 0 column.
Two methods that might be considered are shown below, together with certain
comments. Observe that the numbers in the following displays are those of
the indexing set and only reflect the position of, or number assigned to, each
general Fibonacci sequence.
I. Diagonal Method
1
3
6
10
15
21
28
2
5
9
14
20
27
4
8
13
19
26
7
12
18
25
11
17
24
16
23
22
There are several methods of associating the proper row and column of any
given integer in the above display of positive integers, thus each reader is
free to choose his own favorite scheme. Using the diagonal method of ordering, with S
denoting the n
sequence, we obtain
i070 = (?> ?) = (34,80)
1070 appears in the 35th row and the 12th

column. T 0 + 1 = 35 and Tt - 2T 0 = 12.
(237,475) = S 0
~" ^28411
28411 is found in the 237 + 1 row and the

475 - 2 x 237 column.
Only a short calculation is required to obtain the desired information, n or
S , or (T 0 ,Ti), when given the corresponding data. To retain consistency,
the indexing set is the set of positive integers, (n = l , 2 , 3 , 4 , - )
II. Modified Sides of Squares Method
1
3
7
13
21
31
43
2
4
8
14
22
32
44
5
6
9
15
23
33
45
10
11
12
16
24
34
46
17
18
19
20
25
35
47
26
27
28
29
30
36
48
37
38
39
40
41
42
49
1970]
521
Again, the reader is free to use any one of several well-known methods of
obtaining the row and column of a given integer, and we observe, with S
n
th
denoting the n
sequence of the ordering (n = 1, 2, 3, 4, )
Si070 = (??) = (32,78)
(237,475) = S ? = S5640T
1070 is in the 3 3 I U row and

the 14th column,,
The 2 3 8 m row and the 1 s t
column is the position of
56407.
With regard to the two demonstrated methods of ordering, the diagonal

is possibly a more elegant attack for it parallels that used to count other
infinite matrix arrays.
In conclusion, we have proposed another method of arranging or ordering the general Fibonacci sequences the first being that as suggested by
Brother U. Alfredo
Undoubtedly, there are still others. As implied earlier,
this property of a unique representation of each general Fibonacci sequence

demands the ultimate adoption of some system of ordering to assist the growing number of Fibonacci devotees. References 3 , 4 , 5 , and 6were added by
the Editor.
REFERENCES
1. Brother U. Alfred, ffOn the Ordering of Fibonacci Sequences," Fibonacci
Quarterly, Vol. 1, No. 4, December, 1963, pp. 41-46.
2. Brother U. Alfred, "An Introduction to Fibonacci Discovery," The Fibonacci Association, San Jose State College, California, 1965.
3. P. Naor, "Letter to the Editor," Fibonacci Quarterly, VoL 3, No. 1
(1965), pp. 71-73.
4. Charles H. King, "Conjugate Generalized Fibonacci Sequences," Fibonacci
Quarterly, Vol. 6, No. 1 (1968), pp. 46-49.
5. Eugene Levine,
"Fibonacci Sequences with Identical
Characteristic
Values," Fibonacci Quarterly, VoL 6, No. 3 (1968), pp. 75-80.

6. T. W. Cusick, "On a Certain Integer Associated with Generalized Fibonacci Sequences," ^^^L!^}^!Î^19
126.
Vol. 6, No. 2 (1968), pp. 117-
THE EDUCATIONAL VALUE IN MATHEMATICS

JOHN B. LEWIS
Pasadena, California
The educational value in mathematics is higher to the perceptive individual than the manufacture of things and the mere solving of problems*
I would like to offer a different approach to the significance of the
Golden Ratio, It is for the purpose of making clear to the layman that beyond
the useful pursuit of mathematics for its own sake (which is not an end in itself) lies the deep philosophic content, and a content that constitutes a very
important ingredient of philosophy,,
Plato 1 s Divided Line and Euclid 1 s Golden Section bear identical ratios.
From Plato 1 s exposition, we derive much of the philosophic content,,
Let me begin by noting that the Creative Right Triangle of Pythagoras
(3, 4, 5) is itself created by the encompassing environment of three Right
Triangles whose perpendiculars bear the relation of 2 to 1. This type of right
triangle will be termed,
f!
the Celestial Right T r i a n g l e / ' occupying three-
tenths of a Square* See Figure 1.

From this locating a cosmic position for the creative triangle, it becomes immediately inviting to search for properties of the celestial triangle
which may lead to many analogies,,
We will have more to say about this illustration, which is replete with
Fibonacci and Lucas numbers, as we unfold the picture.
The location of the right angular point of the triangle illustrated in Fig1 lies exactly in the following distances from the boundaries of the square;
South9 1; East 9 2; West, 3; and North, 4 (which is 5 x 5).
When the area of this square is "one," the area of the creative triangle
is three-tenths j and as the cardinal numbers increase, the area of the creative triangle increases in the arithmetical progression of 0.3, 0.6, 0.9, etc.
Therefore, the area of the creative triangle reaches identity with c a r dinal numbers at its intervals in the arithmetical progression of 10 when its
area becomes 3, 6, 9, 12, etc.
When the area of the square is expressed by cardinal numbers squared,
the areas of the creative triangles increase by an increasing progression of
4, 9, 16, 25, etc* The right triangle has sides 3, 4, 5 multiplied by \/E/2B
522
Dec. 1970

N
523
S
Fig. 1 The Celestial Right Triangle
When the areas of the creative triangles reach our squared numbers,
the encompassing square is ten-thirds of them; that i s , if the area of the c r e ative triangle is 36, that square is 120. This indicates a series of Cosmic
squares in the progression of
10
120
403-1/3,
20
30
53-1/3,
163-1/3,
213-1/3,
270
480
563-1/3,
653-1/3,
83-1/3,
333-1/3,
750, etc.
From Figure 1, this means that we have another Cosmic series of

squares, formed with progressive side lengths of y/E9
2VB", 3V5, 4V5 f
etc. , when the square area becomes

5 , 20 ,
45 , 80 , 125,
180 , 245,
320 , 405 , 500, 605, 720,
etc,
The areas of the creative triangle are three-tenths of them, or, in the series
of 6, 24, 54, 96* 150, 216, etc., and therefore in the increasing progression
of 4, 9, 16, 25, etc.
524
[Dec.
In Figure 2, the point of Golden Section is determined at once by taking

the difference between the length of the hypotenuse and the length of the greate r perpendicular, and adding this difference to the length of one-half of this
perpendicular.
Dr. Verner E. Hoggatt, J r . , in his book, Fibonacci and Lucas Numb e r s , ably explains the same situation by a different method of construction.
The square thus divided into these ratio segments, yields in three dimensions, 27 prisms to form the cube.
Among them, only four have variant
volumes.
E
Line Lengths
DC = 1
DA = 0.5
AC = \ / 5 / 2
*AE
= ^ 1 . 1X
2
- V5-2
I ._
g-
__I
__,_
~ -^*s^--
2\/5 + 4
1
AE
D1
1
2
1
X]
_ _
__j
2\/5+4
XY
2
**DE
- *
1
2v^* + 4
9<*
2
1 +V5
-1
= a
=
DX = YC =
V^+3
DY = XC
-1
V5 + 1
XY
1_
- -3
V^ + 2
* Because the Hypotenuse AC minus the greater perpendicular DC equals XC.

**:DE then becomes the length of the Golden Section, equal to DY and XC.
Fig. 2 The point of Golden Section located by Simple Subtraction and Addition
1970]

In F i g u r e 3 9 P l a t o had a t e r m for the l i n e s
EH
and
525
GF,
which he
c a l l e d , "the I n t e l l i g i b l e , "
Line
Lengths
AC =
fi
2\/5 + 4
1?
(EG)
GH
EG
| X
GH
\/5 + 2
(EG)
AB = 1
G
EC = 0.5
G
(GH)
EG X GH
EG X GH
CF =
V5
H
EG
X
GH
(EG)
EG
(EG)
\fE + 3
2
EH =
f
\/5 + 1
E'
The Golden
Ratio
A r e a s of Right T r i a n g l e s C E F and CDF a r e each 0.25,

A r e a s of Golden R e c t a n g l e s
(EH) 2
(EG) 2 =
ABFE
3^+ 7
GEG f G =
GHH ! G f
EEGG 1
\/5 + 2
3V5 + 7
GHHG
5V5 + 11
V5 + 1
\/5 + 3
EEHH 1 =
^5 + 3
%/5+ 1
\EH) 3 =
Cube of Intelligence;
V5 + 1
Cube of M a t h e m a t i c s
(GH)2
(GH)3
17V5 + 38
4V5+9
(EG) 3 =
F i g . 3 I l l u s t r a t i n g the Segmented A r e a s and V o l u m e s
4\/5 + 9
526
[Dec.
It may be observed in Fig. 3 that the cube of the "intelligible" is composed of eight p r i s m s , four only with variant volumes.
-1
4V5 + 9
2
(EG)2 X GH =
12V5 + 29
2
(GH)2 X EG =
21V5 + 47
(EG)3
One
Three
Three
1
=
17V5 + 38
Plato termed the line GH, "Mathematics. ff
(GH)3
Their volumes are:
One
This cube of the "intelligible" has eight positions upon the cube, with
one important feature that they all share in their construction the center
cube, the cube of "mathematics. "
These denominators each share numbers appearing in both the Fibonacci
and Lucas series.
The sum of these eight p r i s m s , that i s , the volume of the cube of the
"intelligible," is
-3
= 2V5 + 4 = V5 + 2
the differenceJêtween atios.

Editorial Note: If
1 + V5
a =
then
thus,
L
=
+ F V5
n
s
1970]
8V5 + 18
13V5 + 29
527
2lV5 + 47
(GH)3 = ~ A _
= a-*
+ 76
_____
___
_._?_^
V. E. H.
Glancing again at Fig 1, an illustration of the "birth" of the creative
triangle, youmayhave noted that the number 1234 is not divisible by "eleven."
See Fig* 4 and accompanying text.
But? suppose we observe these distances in order of rotation
they run as follows:
1243, . 1342,
2134,
2431, 4213,
We find
and 4312, 3124,
3421,
These quaternaries are divisible by 11, and the same holds good for
the remaining possible forty*
Figure 4 illustrates the potential of 48 out of
144 combinations*
The relative proportional areas in Fig* 5 are: - 1 , 4, 4, 6, 5,
or 1,
4, 4, 11. By rotating the triangle into its eight possible positions within the
square, we obtain 24 points which coincide exactly with the points of intersection of perpendicular and horizontal lines within the square of 60 x 60.
By plotting these points, we are provided with the center of the inscribed circle, at x = 0, y = 0; and by bisecting the triangle, we have the
point, x = 0, y = 15 shown in Fig. 5.
Upon making a plotted graph for each of the other triangular positions,
the plotted values of the triangle are merely a matter of sign and number
interchange. The inscribed circle will roll around the circumference of the
circumscribed circle 5 x 0.5 times.
The area of the circumscribed circle
contains the area of the smaller circle 2.5 x 2.5 times,, Area of triangle BDP
is 90. The area of triangle EOF is 20 and 5 / l l t h s .
Multiplying all areas by
11 to clear denominator, we have a total area of xx c 1080 = 11,880. Therefore, the 3,600 square units each enjoy an area of eleven.
Such circumferences increase in the arithmetical progression
3.6X/5 x IT.
Coefficients of\/5
11.30976
22.61952
33.92928
45.23904
56.5488
67.85856
Times Increased
0.5
1.0
1.5
2.0
2*5
3,0
of
528
[Dec.
This illustration gives the numerical evidence at a glance. Any four numbers taken in their rotary position of sequence within the same circle are
divisible by eleven. These 48 numbers, out of a total of 144 possible combinations are unique in their divisibility by eleven. The remaining 96 are
not exactly divisible by eleven.
This chart is not intended to bear any idea of magic, but it does reveal analogies to known Law. The causation of this law is shown in the previous
pages. Philosophic research is just as rewarding as scientific research.
"Ominia numeris sita sunt. " (All things lie veiled in numbers.)
Fig. 4 The Number "Eleven" Chart
1970]
THE
EDUCATIONAL VALUE IN MATHEMATICS
529
A \ x = 30
= 30
x =0
y:
O
^X
'B
~x = 6
y = -18
AREAS
Quadrilaterals
APEF
4725
BDCE
765
Total of the six a r e a s i
Triangles
2475
POE
2475
FEE
PCO
COD
11,800
Fig. 5 The Number "Eleven" Chart

Coefficients of V 5
37.6992
Inscribed Circumference
12 X \/5~ X TT
Increase
56.5488
18 X yjt)
X TT
Circumscribed Circumference
94.248
Factorsj
30 X y/5
4 x 6 x 11 x 17 x 21 x 0.001
= 94.248
X IT
225
1215
530
D e c . 1970
The s u m s and p r o d u c t s of t h e s e quantities yield two quantities in both

n u m e r a t o r s and denominators*
Both Fibonacci and L u c a s n u m b e r s a r e r e p r e s e n t e d in t h e s e n u m e r a t o r s and denominators*
T h i s indicates the home of s p i r a l s making the dynamic
symmetry
within the Golden R e c t a n g l e s ,

The Dual Quantities
3 and
4 and
5 and
11
13 and
29
17 and
38
21 and
47
34 and
76
55 and 123
In the Process of Addition

And Multiplication
987 = (17 x 29) + (13 x 38)
493
+
494
1220 = (21 x 29) + (13 x 47)
609
+
611
2207 = (17V5 x 13VS) + (38 x 29)
1105
+
1102
2728 = (21 Vlf x 13\/5) + (29 x 47),
1365
+
1363
89 and 199
89 and 398
233 and 521
1597 and 3571
ERRATA
P l e a s e m a k e the following c o r r e c t i o n s in "Sums Involving Fibonacci
N u m b e r s , " Vol. 7, No. 1, pp. 92-98:
The f i r s t half of Eq. (3), line 3 , page 9 5 , should r e a d a s follows:
pT
- 2qT
n-1
1
p
+
q
r=0
P l e a s e m a k e the following c o r r e c t i o n s in "Identities Involving G e n e r a l i z e d Fibonacci N u m b e r s / 1 Vol. 7, No. 1, pp. 66-72:
Jv,q)
= 2+
P a g e 67 P l e a s e c o r r e c t line 5 to r e a d :
,,
It i s a l s o e a s y to see that H = p F + q F 1 w h e r e F is the n
Fibonacci
n
n
nx
n
P a g e 69 P l e a s e change the l a s t p a r t of the l a s t sentence of page to read:. .
for the Fibonacci n u m b e r s we get in the g e n e r a l i z e d Fibonacci n u m b e r s the
identity: . . .
P a g e 71 P l e a s e c o r r e c t Eq. (26) to r e a d a s follows:
n
3
H
q>]
[<H!n - q ) + 3e(H2n
x; L-i
r=l
CERTAIN ARITHMETICAL PROPERTIES OF y2k(ak 1)

JOSEPH ARKIN
Spring Valley, New York
Define
= 4 u ( a u - 1)
A
u
B = 4 u < a u
and
&
where a f 0 is a positive rational integer.

In this paper, we discuss
Au +vA
= A,k ,
where the suffixes
Bu+ B v = B,k , ' uA A

v
= A,k
and
BuBv
(u, v, k) are positive rational integers.
= B.k ,
In particular,
we shall, for the first time, finally settle the question, and prove that if one
solution of
û(au l ) | v ( v 1) = - | M k 1)
exists for integral u, v, k, then an infinite number of other such solutions
also exist.
Theorem 1. If a. is an odd integer, then the suffixes are integers in
(2)
-f (a2q2+(2a-l)q+2)
aq+1
+ A
-l(a2q2+2aq-q) '
and
B
-(a2q2+(2a+l)q+2)
where q = 0 , 1 , 2 ,
Proofc
-f (a2q2+2aq+q)
aq+1 '
The proof is immediate, using elementary algebra to show
identities.
Theorem 2. If
(3)
= 4(a 4 q 2 + (2a - l)aq + 2),

531
m = a2q + 1,
532
CERTAIN ARITHMETICAL PROPERTIES OF f k(ak 1)
[Dec.
and
w = (a(n2 + 1) - (n - 1) )/2a ,
then
A =A - + A = A -A
..,
n
n-1
m
w
w-1
(with Hq = 0, 1, 2, ) ,
where the suffixes are integers when a is an odd integer.

Proof. In (2), we replace q with aq and then solve for w in
An = a(n
- 1) = |-(2aw
- a - 1) = Aw - A w-1
- .
2
2
We complete the proof by observing that w and n are integers when a. is
odd.
In the same way we got (3), we get the following:
Corollary. If
n = -f (a4q2 + (2a + l)aq + 2),
m = a2q + 1, and w = (a(n2 + 1) + n - l)/2a ,
then
B = B 1 + B
= B -B
-,
n
n-1
m
w
w-1
(with q = 0, 1, 2, ) ,
where the suffixes are integers when a is an odd integer.

Remark.
It should be noted that R. T. Hansen, in a recent paper [ 1 ] ,
found solutions for the special case when a = 3 in (2), for the A sum, and
in (3).
We now discuss the paired products in the following:
A A = A.
u v
k
and
B B = B, ,
u v
k
for integer suffixes (u, v, k).

Theorem 3. If a is an odd integer, the Pell equation,
1970]
CERTAIN ARITHMETICAL P R O P E R T I E S O F -f k(ak 1)
533
K 2 = 8ap 2 + 8a + 1
(4)
i s solvable in rational i n t e g e r s 9 and K + 1 and 2p 2 + 1 = 0 (mod a ) ,

in A A
= A. , the suffixes
then
(u, v$ k) a r e the following i n t e g e r s
k = (2p2 + l)(2p 3 + 2p 2 + 2p + l ) / a ,
u = (2p 2 + l)(2p 2 + 2p + l ) / a ,
(4.1)
and
i
v = (K + l ) / 2 a = (1 + (8ap 2 + 8a + l ) T ) / 2 a .
Proof.
It i s evident (by e l e m e n t a r y m e a n s ) that the identities in (4.1)
balance the equation A A
= A. .
We complete the proof by noting that the
c o n g r u e n c e s a r e self-evident in (4.1).
T h e o r e m 4.
If a i s an odd i n t e g e r , the P e l l equation

K = 8ap 2 + 8a + 1
(5)
i s solvable in rational i n t e g e r s ,
in B B
= B. , the suffixes
and K - 1 and 4p 2 + 3 = 0 (mod a ) ,
then
(u, v 9 k) a r e the following i n t e g e r s
k = (4p2 + 3)(4p 5 + 4p 4 + 5p 3 + 3p 2 + p ) / a ,
u = (4p2 + 3)(4p 4 + 4p 3 + 3p 2 + p ) / a ,
(5.1)
and
v = (K - l ) / 2 a = (1 + (8ap 2 + 8a + I) 2 ) / 2 a .
Proof.
It i s evident (by e l e m e n t a r y m e a n s ) that the identities in (5.1)
balance the equation B B

U
= B, . The c o n g r u e n c e s in (5.1) of a a r e selfV
evident.
E u l e r [ 2 ] p r o v e d that if
534
CERTAIN ARITHMETICAL P R O P E R T I E S O F | k ( a k 1)
[Dec.
y 2 - Ax 2 = B
(6)
i s solvable in i n t e g e r s , i t s solution r e d u c e s to the integration of the equation

yt+2 - 2myt+1 = 0
in finite d i f f e r e n c e s , the i n t e g r a l being
y = (r + s ) / 2 ,
x = (r - s ) / ( 2 ( A F ) ,
where
r
= (Y + X(A)^)(m + ntA)^) 2 5 " 1 ,
s = (Y - X(A) T )(m - n l A ) ^ ) 2 " 1 ,

Y2 - AX 2 = B ,
Y , X being the l e a s t integral solutions of

the l e a s t integral solutions of
- An
= 1.
and
m,n
being
T h i s i s E u l e r s t h e o r e m in
changed notation.
T h e o r e m 5.
If a i s an odd p r i m e ,
(7)
Kt, P t
(t = 1, 2 , 3 , - . . )
a r e i n t e g e r solutions (where K l 9 P A a r e the l e a s t i n t e g e r solutions) of

</)(t) = K 2 = 8 a P 2 + 8a + 1,
and if t h e r e e x i s t s a K. and a P . which a r e the l e a s t i n t e g e r solutions of
3
</)(t) such that K. + 1 = 0 (mod a) and 2P 2 = - 1 (mod a ) , then the n u m b e r

of solutions of Au A v = A,k a r e infinite for the following& i n t e&g e r suffixes
(u, v , k):
k = (2P 2 + 1)(2P? + 2P 2 + 2 P . + l ) / a ,
l
'
u = (2P? + 1)(2P 2 + 2 P . + l ) / a ,
(7.1)
and
v = OK. + l ) / 2 a = (1 + (8aP 2 + 8a + l ) 2 ) / 2 a
1970]
CERTAIN ARITHMETICAL P R O P E R T I E S OF -ik(ak 1)
w h e r e i = j + w(a - l ) a
ProoL
535
(w = 0S l g 2 , - e ) .
Since K., P . a r e i n t e g e r s , K. + 1 = 0 (mod a) and 2P? + 1 =
0 (mod a ) , then combining (7.1) with (4.1), it i s evident that the

(7.1) a r e i n t e g e r s .
u, v, k
in
Now j combining (6) with the equation 0(j) in (7), we w r i t e

(K. + P.(8a)^)(m + n(8a)^) w a ( a 1 )
3
J
(K. - P . ( 8 a F ) ( m - n(SsF)
j
J
w h e r e a is an odd p r i m e and
In (6) 3 it i s evident that
R[
w=0sl
= K
= K
2,
, -v^. + P
. ^.(Sa)2" ,
wa(a-l)+]
wa(a-l)+jx
, 1U. - P
, -.^-(Sa) 2 ,
wa(a~l)+j
wa(a-l)+jx
a 8 O
(m,a) = 1, and since a i s an odd p r i m e 9 we
h a v e , by F e r m a t s f a m i l i a r t h e o r e m (m, a a r e i n t e g e r s with a an odd p r i m e ,

a-1
(m, n) = 1, then m
= 1 (mod a) )
(m n ( 8 a F ) w a ( a " 1 )
= 1 (mod a) ,
which l e a d s to (in (8)),

K. = K
, ^ .
j
w a ( a - l )Xx
+]
(mod a)
and
P. = P
, 1U. ,
j
wa(a~l)+j
and we complete the proof by noting that t h e s e c o n g r u e n c e s satisfy the c o n ditions of T h e o r e m 5.

C o r o l l a r y 1.
(9)
Since 9
In (7), it i s a l m o s t i m m e d i a t e that
1 < j < a(a - 1)
if j = sa(a - 1) + d (where 1 = d = a (a - 1) and a = 0 5 1, 2 , ) ,
it i s evident that
K
/ - v ,. = K / 1 W , \ , ,
wa(a-l)+j
a(a-l)(w+s)+d
= K,
d
(mod a ) ,
and
P
, -v,. = P / 1W , \ , J =
wa(a-l)+j
a(a-l)(w+s)+d
PJ
N(mod
a)
536
CERTAIN ARITHMETICAL PROPERTIES OF | k ( a k 1)
[Dec.
Corollary 2. If a is an odd prime, and

K. - I s
0 (mod a)
and
then the number of solutions of B B

integer suffixes (u, v, k):
4P 2 + 3 = 0 (mod a) ,
= B,
are infinite for the following
k = (4P2 + 3)(4PJ| + 4Pi + 5Pj + 3 P | + P i ) / a ,

u = (4P* + 3)(4Pj + 4Pi + 3Pi + P ^ / a ,
and
i
v = (K. - l)/2a = (1 + (8aPi + 8a + l) T )/2a ,

where
i = j + wa(a - 1)
(w = 0, 1, 2, ) ,
and
1 < j < a(a - 1) .
We shall give one application in pentagonal numbers for infinite paired
products in (7-7.1).
In (7-7.1), let a - 3, then
K2
= 24P 2 + 25
and
m 2 = 24n2 + 1,
where the first solutions are

Kj = 7,
Pj = 1,
and
m = 5,
n = 1.
We then find that j = 4 and 6, so that i = 6w + 4 and 6w + 6, and we write

(7 + (24)*)<5 + (24)^) 6 w + 3 = (K4 + P4(24)^)(5 + (24)*) 6w = K g w + 4 + P 6 w + 4 (24)*= r ,
1970]
CERTAIN ARITHMETICAL P R O P E R T I E S O F | k ( a k + 1)
537
and
( 7 - (24)*)(5- ( 2 4 ) * ) 6 w + 3 = (K4 - P 4 ( 2 4 ) * ) ( 5 - (24)*) 6 w = K g w + 4 - P 6 w + 4 ( 2 4 ) * = s ,
so that
find
(r + s ) / 2 = K fi
(r + s ) / 2 = K fi
fi
+4
and
(r - s ) / 2 = P g
and (r - s ) / 2 = P
g.
+4,
In the s a m e way, we
Then combining t h e s e
re-
s u l t s with (6) and (7.1) 5 we conclude o u r application.

REFERENCES
lo
R. T. H a n s e n , " A r i t h m e t i c of Pentagonal N u m b e r s , " Fibonacci Q u a r t e r l y , Vol. 8, No. 2 (1970), pp. 83-87.
2.
L. E u l e r , C o m m , A r i t h . Coll. , I, pp. 316-336.
[Continued from page 475. ]
4).
3,
H e n c e , if odd p r i m e p divides F 2 -
1?
then p i s not of the f o r m 4s +
thus proving Conjecture 2 of D m i t r i T h o r o . * The proof by L e o n a r d Wein-
stein**
c a m e to m y attention a t a l a t e r t i m e and i s d i s t i n c t from the above
proof.
* D m i t r i T h o r o , "Two Fibonacci C o n j e c t u r e s , " Fibonacci Q u a r t e r l y ,
1965, pp. 184-186.
** L e o n a r d Weinstein,
1966, p . 88.
" L e t t e r to the E d i t o r , " Fibonacci Q u a r t e r l y ,

ô-ô^
Oct.
Feb.
ERRATA
P l e a s e m a k e the following c o r r e c t i o n s in ? Some R e s u l t s on Fibonacci
Q u a t e r n i o n s , " Vol. 7, No. 2, pp. 201-210.
P a g e 201 The f i r s t displayed equation on the page should r e a d :
j2 = j2 = k 2 = _ 1 ? y = _ji = k ; j k = _ k j = i ; k i = _ i k = ^
P a g e 205 Change the b r a c k e t e d p a r t of Eq. (27) to r e a d :
[FrT0 + F2r(Q0 " 3k)l
P a g e 208 Change the f i r s t t e r m s of Eq. (74) to r e a d :
T
F
n+t n+r
FIBONACCI NUMBERS
AS PATHS OF A ROOK ON A CHESSBOARD
EDWARD T. FRANKEL
Schenectady, New York
The purpose of this article is to show that Fibonacci numbers can be

derived by enumerating the number of different routes of a rook from one
corner of a chessboard to the opposite corner when the moves of the rook are
limited by restrictive fences.
Consider the chessboard array of binomial coefficients orfigurate numbers in Figo 1. It is well known that the number in any square or cell r e p r e sents the number of different routes of a rook from the upper left corner to
that cell, provided that the rook moves are either horizontal to the right or
vertically downward.,*
1
10
15
21
28
36
10
20
35
56
84 120
15
35
70 126 210 330
21
56 126 252 462 792
28
84 210 462 924 1716
36 120 330 792 1716 3432
Fig* lo Number of Rook Paths from Corner of Chessboard
Figure 2 shows the same chessboard array in standard combinatorial

notation:
(k) = WkMh-k)! = ( h h _ k )
*EdouardLucas, Theorie des Nombres, P a r i s , 1891, page 83.

538
FIBONACCI NUMBERS AS PATHS

OF A ROOK ON A CHESSBOARD
Dec. 1970
539
(?) (?) (?) (?) (J) (?) (?)

(?) (?) ( ! ) (?) (?) (?)
(?) II) (1) (5) (?) (?) (?)
(?) (?) (?) (?) (?) (?)
(1) (?) 0 (I) (?) (?) (?)
(?) (?) (I) (?) (?) (?) (?)
(?) (?) (?) (?) (?) (?) (?)
(?) (?) (?) (?) (?) (?) (?)
(?)
(?)
G)|
CO CO
Fig. 2.
(?)
(?)
(?)
(?)
(?)
Rook P a t h s in Combinatorial Notation
F i g u r e 3 shows a c h e s s b o a r d a r r a y w h e r e the m o v e s of a rook a r e

l i m i t e d by the indicated p a t t e r n of horizontal and v e r t i c a l r e s t r i c t i v e fences.
13
21
21
13
34
55
34
89
144 144
89
233 377
55
233 610
F i g . 3 Rook P a t h s L i m i t e d by R e s t r i c t i v e F e n c e s
The a r r a y begins with n u m b e r one in the top left corner*
Inasmuch a s the
n u m b e r in any cell i s the s u m of the n u m b e r s i m m e d i a t e l y above it and to the

left of it 5 the p a t t e r n of r e s t r i c t i v e fences r e s u l t s in the e n t i r e a r r a y being
composed of Fibonacci n u m b e r s .
[Dec.
FIBONACCI NUMBERS AS PATHS
540
Figure 4 shows the chessboard with the same pattern of fences as in

Fig. 3, but with the numbers in the Fibonacci notation where F 0 = Ft = 1;
F 2 = 2; F 3 = 3; and, in general, F.
"n+2
F _L1 + F .
n+1
n
1 F
Fi
Fi
F0
F2
F3
F3
F2
F4
F5
F5
F4
F6
F7
F7
F6
F8
F9
F9
F8
10
Fn F
10
12
i2
FU
13 J
Fig. 4. Limited Rook Paths in Fibonacci Notation

Comparing Fig. 4 with Fig. 2, it is noted that F 0 corresponds to ( n J
Il4t\
and Fj4 corresponds to 1 7 J. Comparing Fig. 3 with Fig. 1, we see that
the number of Fibonacci rook paths from corner to corner is 610, whereas
the number of unrestricted paths is 3432. The difference of 2822 must be the
number of routes which are eliminated because of the restrictive fences. This
can be verified by tabulating the effect of each restrictive fence as in the analysis on the following page.
To generalize, in a chessboard of (n + I) 2 cells, the number of unrestricted rook paths from corner to corner is
J , the number of Fibonacci
rook paths is F 2 , and the number of paths that are eliminated by the pattern of horizontal and vertical fences is
(?)
2n
* . (
\ -
+ F 2n-6 4
( s)
+F
Jt)
' . ( " % -
+
+F
2n-5
(S)
19701
OF A ROOK ON A CHESSBOARD
541
ANALYSIS OF ELIMINATED ROOK PATHS

B
Number of Fibonacci paths
from origin to cells with
fences
Number of unrestricted
rook paths from fence
to lower right corner
Ax B
Number of paths
eliminated by
fences
Cells with horizontal

fences
=
F2
F4
Fe
= 13
F8
= 34
F 1 0 = 89
ra
G)
ra
(t)
= 792
792
= 210
420
56
280
15
195
136
89
it)
to to
Fo
Subtotal
1912
Cells with vertical

fences
(?) -
330
Fi =
F3 =
F5 =
(I)
168
F 7 = 21
(a
105
F 9 = 55
( ! ) -
262
( ! )
"
55
*
Subtotal
liminated paths
Ibonacci paths, F ^
Total number of unrestricted paths
910
2822
ft
610
3432
ELEMENTARY PROBLEMS AND SOLUTIONS

Edited by
A. P.HILLMAN
University of New Mexico, Albuquerque, New Mexico
Send all communications regarding Elementary Problems and Solutions

to Professor A. P. Hillman, Dept. of Mathematics and Statistics, University
of New Mexico, Albuquerque, New Mexico 87106.
Each problem or solu-
tion should be submitted in legible form, preferably typed in double spacing,

on a separate sheet or sheets, in the format used below.
Solutions should
be received within three months of the publication date.

Contributors (in the United States) who desire acknowledgement of r e ceipt of their contributions are asked to enclose self-addressed
stamped
postcards.
B-196
Proposed by R. M. Grassl, University of New Mexico, Albuquerque, New Mexico.
Let a 0 , a l 9 a 2 , , and b 0 , b*, b 2 ,
n = (S)an
(l) a n-l
Give the formula for a
B-197
V2
s88
be two sequences such that
(n)a0
n = 0, 1, 2,
in terms of b , 8 , b A .
n
n5
' 0
Proposed by Phil Maria, University of New Mexico, Albuquerque, New Mexico.
Let the Pell Sequence be defined by P 0 = 0,

2P
- + P . Show that there is a sequence Q
n+2k
k P n+k " ("1}
P 4 = 1,
such that
'
and give initial conditions and the recursion formula for Q .

542
and P
Dec. 1970
B-198
543
Proposed by Phil Mana, University of New Mexico, Albuquerque, New Mexico.
be the coefficient of xtx2 e e x
Let c
in the expansion of
(-Xj + X 2 + X3 + . - + X n ) ( x 1 - X2 + X3 + + X n ) ( x 1 + X 2 - X 8 + + X n ) o e .
(x1 + x 2 +x 3 + + x n i - x n ) .
For example s ct = - 1 , c 2 = 2, c 3 = - 2 , c 4 = 8, and c 5 = 8B
c
l0
n+2
= nc ,- + 2(n + l)c ,
n+1
n
Show that
c = nc . + (-2) s,
n
n-1
and
lim (c /nl) = e
n *- oo n /
B-199
-2
Proposed hy M. J. DeLeon, Florida Atlantic University, Boca Raton, Florida.
Define the Fibonacci and Pell numbers by

Fi = 1, F 2 = 1, F ^ = F _,, + F
1
A
$
'
n+2
n+1 n
Pi = 1, P 2 = '2, P _,0 = 2P . + P
1
s
L
$
n+2
n+1 n
Prove or disprove that P g , < F - - ,
B-200
n>1 ;
n> 1.
for k > 1.
Proposed by M. J, DeLeon, Florida Atlantic University, Boca Raton, Florida.
With the notation of B-199, prove or disprove that
B-201
llk <
6k+1
f 0 r k
Proposed by Mel Most, Ridgefield Park, New Jersey.
Given that a very large positive integer k is a term F
in the Fib-
onacci Sequence, describe an operation on k that will indicate whether n

is even or odd*
544
[Dec.
SOLUTIONS
DOUBLING NEED NOT BE TROUBLING
B-l 78
Proposed by James E. Desmond, Florida State University, Tallahassee, Florida.
For all positive integers n show that

F
+ z
2n+2 - E ^ " V i
"
i=l
and
n
F
2n+3
2n
"F2i+
2n+1
i=l
Generalize.
Solution by Herta T. Freitag, Hollins, Virginia.
Our generalization states that for all positive integers n and for all
positive integers a
<
2 lF
" 2i + (a-3)
+ 2
= F
2n + a
i=l
The proof is by mathematical induction on n.
Relationship (1), for n = 1, claims that
F
- + 2F = F ^ ,
a-1
a
a+2
which i s , indeed, the case

Assuming that (1) holds for some positive integer, say k, we have:
1970]
545
< 2)
Z ^Sâ-S)
+2
\ = *2k+a
1=1
To see if
k+1
i=l
we recognizeon the basis of our assumption (2)that the left side of (3)
equals
2F
2k+a
2k+a-l
whichj howeverj is seen to be the number

F
F ,
+ F~,
- and, hence,
2k+a+2
This completes our proof by the principle of mathematical induetione
The relationships stated in the problem now become special
ases of
our generalization (1), whereby a = 2 establishes the first, and a = 3 the

second, of the two given formulas.
Also solved by C. B. A. Peck, A. G. Shannon (T.P.N.G.), and the Proposer.
A SURJECTION (NOT MONOTONIC)

B-l 79
Based on Douglas Lind's Problem B-165.
Let Z
Z
to Z
4.
consist of the positive integers and let the function b from
be defined by b(l) = b(2) = 1,
b(k + 1) +b(k) for k = 1, 2,
eso
b(2k) = b(k),
and b(2k + 1) =
. Show that every positive integer m is
a value of b(n) and that b(n + 1) > b(n) for all positive integers n.
Solution
Put
546
[Dec.
b(k)x]k-1
B(x) = J^
k=l
Then
00
2k-2
2k 1
B(x) = J^ b(2k)x " + ] T b(2k - l)x :

k-1
k=l
00
00
b(k)
x2k_1 + 1 +
k=l
[b(k) * b(k + 1 ) ] x 2 k
k=l
= xB(x 2 ) + x2B(x2) + ] T b(k + l ) x 2 k ,

k=0
so that
B(x) = (1 + x + x 2 )B(x 2 )
It follows that
B(x) = J ! (1 +
n=0
x2n
+ *2
It is evident from this generating function that every positive integer m is a

value of b(n). However, the statement b(n + 1) > b(n) is false:
b(2k + 2) = b(k + 1) < b(k + 1) + b(k) = b(2k + 1) .
For additional properties of b(k), see;
?f
A Problem in Partitions Re-
lated to the Stirling N u m b e r s / ' Bull. Amer. Math. Soc. , Vol. 70 (1964), pp.
275-278; also: D. A. Lind, ffAn Extension of Stern 1 s Diatomic S e r i e s / ' Duke
Math. Journal, Vol. 36 (1969), pp. 53-60.
1970]
547
Outline
Every positive integer m is a value of b(n) since
m = b ( 2 m _ 1 + 1) .
This is easily established by mathematical induction using the definition of
b(n) and the fact that b(2 n ) = 1.
BUNNY PATHS?
B-180
Proposed by Reuben C. Drake, North Carolina A & T University, Greensboro, North Carolina.
Enumerate the paths in the Cartesian plane from (0,0) to (n,0) that
consist of directed line segments of the four following types:
|j
Type
Initial Point
T e r m i n a l Point
|1
II
III
IV
(k,0)
(k,0)
(k,l)
(k,l)
(k.l)
(k + 1,0)
(k + 1,1)
(k + 1,0)
Solution by L. Carlitz, Duke University, Durham, North Carolina.
Let f(n) denote the total number of paths from (0,0) to (n s 0) e
Let
f0(n) denote the number of paths ending with segment of Type II, and fj(n)
the number ending with a segment of Type IV. Then we have
f0(n + 1) = f0(n) + fi(n) = f(n)
fi(n + 1) = f (0) + f(1) + + f(n) .
It follows that
n
f(n + 1) = f(n) + ^2 f(k) .
k=0
Put
548

00
F(x) = J2 f(n)xn .
n=0
Then
oo I
F(x) = 1 + J ] ]f(n) + J2 f(k)[xn

n=0 I
k=0
= 1 + xF(x) + jXFix)
so that
F(x) =
* ~ X
1 - 3x + x2
Since
CO
1 - 3x + x2
= I>2nX
n=0
in the usual notation for Fibonacci numbers, it follows that

f(n) = F
2n+l
Moreover,
fo(n) =
F ^
and
f!(n) = F 1
F8 + + F 2 n _ 1 = F 2 n
Also solved by G. J. Giaccai and the Proposer.
[Dec.
1970]
549
AN INFINITE MATRIX
B-181
Proposed by J. B. Roberts, Reed College, Portland, Oregon.
L e t m be a fixed i n t e g e r and l e t G - = 0,
&
-1
for n > 1.
Show that GQS G
, G2
, G.
Gi = 1,
*
G = G
n
- +G
n_i
n_2
i s the sequence of u p p e r
left p r i n c i p a l m i n o r s of the infinite m a t r i x
m-2
G 0 + G
m-2
m
(-Dm
m-2
(-Dm
G
m--2
>
(-D
Expansion of the typical m i n o r M, , k > 2 S by m e a n s of the e l e m e n t s

of i t s l a s t row j yields the r e c u r r e n c e r e l a t i o n
M.k = (Gm 2o + Gm )M.k - 1- - ( - D M , k -Q2
Induction 9 m a k i n g u s e of the identity
G
= (G, 0 + G, )G , - (-1) G of
k-2
k n-k
n-2k
(itself e a s i l y p r o v e d by induction), yields the conclusion,

CONGRUENCES
B-l 82
Proposed by James E. Desmond, Florida State University, Tallahassee, Florida.
Show that for any p r i m e p and any i n t e g e r n 9

F
np
= F F N(mod Fp)
n p
and
L
np
= L L
n p
s L N(mod p)
n
^
550
[Dec.
We h a v e , from H a r d y and Wright, T h e o r y of N u m b e r s , Oxford U n i v e r sity P r e s s , London, 1954, p 0 150, that F

p) if p = 1 (mod 5) and that F
p = 2 (mod 5).
Therefore,
- = 0 (mod p) and F
- = 0 (mod p)
and
= 1 (mod
s - 1 (mod p)
== 1 (mod p) for all p r i m e s p ,
if
the c a s e
p = 5 being c l e a r .
F r o m I. D. R u g g l e s ,
Some Fibonacci R e s u l t s Using F i b o n a c c i - T y p e
Sequences, ' Fibonacci Q u a r t e r l y , Vol. 1, No. 2 (1963), p . 77, we have

F _,_ = L F + (-1)
r+s
s r
for all i n t e g e r s r and
a r b i t r a r y i n t e g e r n
s.
Let
s = p,
r-s
a p r i m e , and let
r = np
for an
Then
F , ^-v
= L F
+ x( - 1 ) P + 1 F / -v
(n+l)p
p np
(n-l)p
so t h a t
F>(n+l)p
^-n =
np
for any i n t e g e r n and any p r i m e p ,
/ (n-l)p
t\
(Nm o d p^)
the c a s e p = 2 being c l e a r .
Similar-
l y , we obtain
L, , - v
(n+l)p
E L
np
+ L / - v v (mod p)
(n-l)p
^
for any i n t e g e r n and any p r i m e p , since L

gers m.
Now, using induction on n ,
Suppose F
F
= F F
/ (k+l)p
iî\v,=
Suppose F
ikp
= F F
= F
- +F
., for all i n t e -
the proposition i s t r u e for n = 0, 1.
( m o d p ) for n = 0, 1, . . . , k with k > 0.

+
/ i(k-l)p
i\
= F .kF p
Then
F. k --1F p ( m o d p ^) .
( m o d p ) for n = 1, 0, , k with k > 1.
Then
F/, t\
= F / f , - , - F.
s F , _ , - F - F. F
= F. - F N( m o dFp ) .
(k-l)p
(k+l)p
kp
k+1 p
k p
k-1 p
1970]
551
Therefore 5 the Fibonacci congruence relation is true for any prime p and
any integer n8 The Lucas congruence relation can be proved by an argument similar to that given above.
PALINDROME CUBES
B-183
Proposed by GustavusJ. Simmons, Sandia Corporation, Albuquerque, New Mexico.
A positive integer is a palindrome if its digits read the same forward

or backward,
The least positive integer n9 such that n2 is a palindrome
but n is not ? is 26e Let S be the set of n such that n3 is a palindrome

but n is note Is S empty, finite, or infinite?
Comment by the Proposer.
Since 22013 is the palindrome 10662526601, S is not empty. This is

all that is known about the set S*
[Continued from page 506. ]

a = 29
30
b =
35
113
c =
48
113
31
97
120
32
65
65
33
34
65
34
145
145
35
73
102
36
61
65
37
37
70
38
181
181
39
41
50
40
101
101
VOLUME INDEX
ANDERSON, CHRISTINE.
P r o b l e m Solved: B - 1 6 4 , Vol. 8, No. 1, p. 111.
ANDERSON, JEAN H. " S p i r a l s , C h e c k e r b o a r d s , P o l y o m i n o e s , and the F i b onacci S e q u e n c e , " Vol. 8, No. 1, pp. 90-95.
ANDREWS, GEORGE E.
P r o b l e m P r o p o s e d : H - 1 3 8 , Vol. 8, No. 1, p . 76.
ANGLIN, ROBERT H. P r o b l e m P r o p o s e d : B - 1 6 0 , Vol. 8, No. 1, p. 107.

ARKIN, JOSEPH.
" R e c u r r e n c e F o r m u l a s , " Vol. 8, No. 1, pp. 4 - 5 (coa u t h o r , R i c h a r d Pollack). "An Extension of Fibonacci N u m b e r s - I I , "
Vol. 8, No. 2 , pp. 199-216 (co-author, V. E. Hoggatt, J r . ) . "An
Extension of a T h e o r e m of E u l e r , " Vol. 8, No. 4 , pp. 421-427.
" C e r t a i n A r i t h m e t i c a l P r o p e r t i e s of ^k(ak + 1 ) , " Vol. 8, No. 5, pp.
531-537.
BARLEY, W. C. P r o b l e m s Solved: B - 1 7 2 , Vol. 8, No. 4 , p. 445; B - 1 7 5 ,
Vol. 8, No. 4 , p. 445.
BEVERAGE, DAVID G. " L e t t e r to the E d i t o r , " Vol. 8, No. 5, p. 498. "On
a Conjecture of D m i t r i T h o r o , " Vol. 8, No. 5, p. 475.
BICKNELL, MARJORIE. "Summation of P o w e r s of Roots of Special E q u a t i o n s , " Vol. 8, No. 2, pp. 221-224 ( c o - a u t h o r , N. A. D r a i m ) . "A
P r i m e r for the Fibonacci N u m b e r s : P a r t V I I , " Vol. 8, No. 4 , pp. 4 0 7 420.
BRADY, W R A Y G . P r o b l e m s Solved: B - 1 6 6 , Vol. 8, No. 3 , p . 327; B - 1 7 2 ,
VoL 8, No. 4 , p. 445; B - 1 7 3 , Vol. 8, No. 4 , p . 445; B - 1 7 5 , Vol. 8,
No. 4 , p. 447; H - 1 5 5 , Vol. 8, No. 5, p. 498.
BROWN, J . L. , J R . "Modulo One Uniform Distribution of the Sequence of
L o g a r i t h m s of C e r t a i n R e c u r s i v e S e q u e n c e s , " Vol. 8, No. 5, pp. 4 8 2 486 (co-author, R. L. Duncan). P r o b l e m P r o p o s e d : H-137, VoL 8,
No. 1, p . 75. P r o b l e m Solved: H - 1 3 7 , Vol. 8, No. 1, p. 75.
BROUSSEAU, BROTHER A L F R E D . " L i n e a r R e c u r s i o n Relations L e s s o n
Seven, Analyzing L i n e a r R e c u r s i o n R e l a t i o n s , " VoL 8, No. 1, pp. 9 6 101. "Book R e v i e w , " Vol. 8, No. 3 , p. 280. " L i n e a r R e c u r s i o n R e lations L e s s o n Eight, Asymptotic Ratios in R e c u r s i o n R e l a t i o n s , "
VoL 8, No. 3 , pp. 311-316. "A Lucas A n a l o g u e , " VoL 8, No. 4 ,
pp. 439-442.
BRUCKNER, GOTTFRIED. " F i b o n a c c i Sequence Modulo A P r i m e ,
(mod 4 ) , " Vol. 8, No. 2 , pp. 217-220.
552
P = 3
D e c . 1970
BURNS, DANIEL W.
BUSCHMAN, R. G.
pp. 372-374.
VOLUME INDEX
Problem Proposed:
553
H-174, Vol. 8, No. 4 , p. 383.
' T h e 'Difference S e r i e s ' of M a d a c h y , " Vol. 8, No. 4 ,
C A R L I T Z , LEONARD. "Some Fibonacci and L u c a s I d e n t i t i e s / ' Vol. 8, No.

1, pp. 61-73 ( c o - a u t h o r , H. H. F e r n s ) . " F i b o n a c c i R e p r e s e n t a t i o n s
2 , " Vol. 8, No. 2 , pp. 113-134. "Some G e n e r a l i z e d Fibonacci Identit i e s , " Vol. 8, No. 3 , pp. 249-254. P r o b l e m s P r o p o s e d : H - 1 6 7 , Vol.
8, No. 1, p. 74; H - 1 3 9 , Vol. 8, No. 3 , p. 269; B - 1 8 5 , Vol. 8, No.
3 , p. 325; B - 1 8 6 , Vol. 8, No. 3 , p. 326; H - 1 4 4 , Vol. 8, No. 4 , p.
384; H - 1 7 5 , Vol. 8, No. 5, p. 487; H - 1 7 7 , Vol. 8, No. 5, p. 488;
H - 1 5 1 , Vol. 8, No. 5, p. 488; H-154, Vol. 8, No. 5, p. 493.
P r o b l e m s Solved: H - 1 3 9 , Vol. 8, No. 3 , p. 269; H-142, Vol. 8, No.
3 , p . 275; B - 1 6 9 , Vol. 8, No. 3 , p. 330; H-1 44, Vol. 8, No. 4 , p.
385; H-149, Vol. 8, No. 4 , p. 391; B - 1 7 6 , Vol. 8, No. 4, p. 448;
B - 1 7 7 , Vol. 8, No. 4 , p. 448; H - 1 5 1 , Vol. 8, No. 4 , p. 492; H - 1 5 4 ,
Vol. 8, No. 5, p. 493; B - 1 7 9 , Vol. 8, No. 5, p. 545; B - 1 8 0 , Vol. 8,
No. 5, p. 547.
CARLSON, JOHN R.
pp. 499-506.
" D e t e r m i n a t i o n of Heronian T r i a n g l e s , " Vol. 8, No. 5,
CHEVES, WARREN. P r o b l e m s Solved:

B - 1 9 2 , Vol. 8, No. 4 , p. 443.
B-168,
Vol. 8,
No. 3 , p. 328;
CHURCH, C. A. , J R . "On the E n u m e r a t i o n of C e r t a i n T r i a n g u l a r A r r a y s , "

Vol. 8, No. 3 , pp. 2 3 5 - 2 4 1 .
CLARKE, ED. P r o b l e m Solved: B - 1 6 8 , Vol. 8, No. 3 , p. 329 (co-solved by
Martha Clarke).
CLARKE, MARTHA. P r o b l e m Solved: B - 1 6 8 , Vol. 8, No. 3 , p. 329 (cosolved by Ed C l a r k e ) .
COHN, ERNST M. "Complete Diophantine Solutionof the P y t h a g o r e a n T r i p l e , "
Vol. 8, No. 4 , pp. 402-405.
COX, NANNETTE. "Some U n i v e r s a l C o u n t e r e x a m p l e s , " Vol. 8, No. 3 , pp.
242-248 ( c o - a u t h o r s , John W. Phillips and V. E. Hoggatt, J r . ) .
CRENSHAW, JOE R. "Some R e m a r k s on the O r d e r i n g of G e n e r a l Fibonacci
S e q u e n c e s , " Vol. 8, No. 5 , pp. 516-521.
C U L L E N , T. J . P r o b l e m s Solved: B - 1 7 2 , Vol. 8, No. 4 , p. 445; B - 1 7 3 ,
Vol. 8, No. 4 , p. 445; B - 1 7 4 , Vol. 8, No. 4 , p. 446; B - 1 7 5 , Vol. 8,
p. 448.
DAVENPORT, J U L I E T T E .
VOLUME INDEX
554
[Dec.
DAYKIN, D. E. T f a c t o r i z a t i o n of Fibonacci N u m b e r s , " Vol. 8, No. 1, pp.

23-30 (co-author, L. D r e s e l ) . "An Addition Algorithm for G r e a t e s t
Common D i v i s o r / ' V o l . 8, No. 4 , pp. 347-349.
DE CARLI, D. J. n A G e n e r a l i z e d Fibonacci Sequence Over an
R i n g , " Vol. 8, No. 2 , pp. 182-184.
DE KOVEN, FRANCIS.
Problem Proposed:
Arbitrary
H-169, Vol. 8, No. 3, p. 268.
DE LEON, MORRIS JACK. " T h e Fibonacci N u m b e r s C o n s i d e r e d a s a P i s o t

S e q u e n c e , " Vol. 8, No. 5, pp. 4 7 6 - 4 8 1 . P r o b l e m s P r o p o s e d : B - 1 9 9 ,
DESMOND, JAMES E. P r o b l e m s P r o p o s e d : B - 1 7 8 , Vol. 8, No. 1, p. 105;
B - 1 8 2 , Vol. 8, No., 1, p. 106; B - 1 7 8 , Vol. 8, No. 5, p. 544; B - 1 8 2 ,
Vol. 8, No. 5, p. 549. P r o b l e m Solved: B - 1 8 2 , Vol. 8, No. 5, p. 550.
DIAS, ANDREW.
DRAKE, REUBEN C. P r o b l e m P r o p o s e d : B - 1 8 0 , Vol. 8, No. 1, p. 106;

B - 1 8 0 , Vol. 8, No. 5, p. 547. P r o b l e m Solved: B - 1 8 0 , Vol. 8, No. 5,
p. 548.
DRAIM, N. A. "Summation of P o w e r s of Roots of Special E q u a t i o n s , " Vol,
8, No. 2, pp. 221-224 (co-author, M a r j o r i e Bicknell).
DRESEL, L. A. G. " F a c t o r i z a t i o n of Fibonacci N u m b e r s , " Vol. 8, No. 1,
pp. 23-30 (co-author, D. E. Daykin).
DUNCAN, R. L. "Modulo One Uniform Distribution of the Sequence of L o g a r i t h m s of C e r t a i n R e c u r s i v e S e q u e n c e s , " Vol. 8, No. 5, pp. 482-486
(co-author, J. L. Brown, J r . ).
ENGLUND, DAVID. P r o b l e m P r o p o s e d : H-172, Vol. 8,. No. 4 , p. 383.
P r o b l e m s Solved: B - 1 7 2 , Vol. 8, No. 4, p. 445; B - 1 7 4 , Vol. 8, No.
4 , p. 446.
FERGUSON, DAVID E. " L e t t e r s to the E d i t o r , " Vol. 8, No. 1, pp. 88-89
(co-author, H a r l a n Umansky).
FERNS, H. H. "Some Fibonacci and L u c a s I d e n t i t i e s , " Vol. 8, No. 1, pp.
61-73. P r o b l e m P r o p o s e d : H-166, Vol. 8, No. 1, p. 74.
FINKELSTEIN, R A P H A E L , "Application of R e c u r s i v e Sequences t o D i o p h a n tine E q u a t i o n s , " Vol. 8, No. 5, pp. 463-469 ( c o - a u t h o r s , E d g a r K a r s t
and Hymie London).
FRANKEL, EDWARD T. " F i b o n a c c i N u m b e r s a s P a t h s of a Rook on a C h e s s b o a r d , " Vol. 8, No. 5, pp. 538-541.
1970]
VOLUME INDEX
555
F R I E T A G , HERTA T.
P r o b l e m s Solved: B - 1 6 0 , Vol. 8, No. 1, p. 107;
B - 1 6 1 , Vol. 8, No. 1, p. 108; B - 1 6 3 , Vol. 8, No. 1, p. 110; B - 1 6 4 ,
Vol. 8, No. 1, p. I l l ; B - 1 6 5 , Vol. 8 S No. 1, p. 112; B - 1 6 8 , Vol. 8,
No. 3, p. 328; B - 1 7 2 , Vol. 8, No. 4, p. 445; B - 1 7 3 , Vol. 8, No. 4,
p. 445; B - 1 7 4 , Vol. 8, No. 4 , p. 446; B - 1 7 5 , Vol. 8, No. 4 , p. 447;
B - 1 7 6 , Vol. 8, No. 4 , p. 448; B - 1 7 7 , Vol. 8, No. 4 , p. 448; B - 1 7 8 ,
Vol. 8, No. 5, p. 544.
GABAI, HYMAN.
31-38.
GIACCAI, G. J.
" G e n e r a l i z e d Fibonacci k - S e q u e n c e s , " Vol. 8, No. 1, pp.

G L A Z E , JANET WATERMAN. "Angle Multisection by P a r a l l e l

e d g e s , " Vol. 8, No. 4 , pp. 393-396.
Straight-
GOULD, H. W. P r o b l e m s P r o p o s e d : H-170, Vol. 8, No. 3 , p. 368; H - 1 4 2 ,

Vol. 8, No. 3 , p. 275.
GRASSL, R. M. P r o b l e m P r o p o s e d : B - 1 9 6 , Vol. 8, No. 5, p. 542.
l e m Solved: B - 1 7 7 , Vol. 8, No. 4, p. 448.
Prob-
GRONEMEIJER, CARL.
P r o b l e m P r o p o s e d : B - 1 8 7 , Vol. 8, No. 3 , p. 326.
G U I L L O T T E E , GUY A.
P r o b l e m P r o p o s e d : B - 1 9 1 , Vol. 8, No. 4, p. 443.
GUPTA, A. K. P r o b l e m s Solved:
Vol. 8, No. 3 , p. 333.
B - 1 7 0 , Vol. 8, No. 3 , p. 332;
B-171,
HAGIS, P E T E R , JR. rTA T h e o r e m Concerning Odd P e r f e c t N u m b e r s , " Vol.

8, No. 4 , pp. 337-346 (co-author, D. S u r y a n a r a y a n a ) .
HAMELIN, S.
P r o b l e m Solved: H - 1 5 0 , Vol. 8, No. 4 , p. 392.
HANSEN, RODNEY T.
pp. 83-87.
" A r i t h m e t i c of Pentagonal N u m b e r s , " Vol. 8, No. 1,
HARRIS, V. C. "An Algorithm for Finding the G r e a t e s t Common D i v i s o r , "

Vol. 8, No. 1, pp. 102-103. "Note on the N u m b e r of Divisions R e quired in Finding the G r e a t e s t Common D i v i s o r , " Vol. 8, No. 1, p.
104.
HAUTUS, M. L. J. P r o b l e m Solved:
solved by D. A. K l a r n e r ) .
H - 1 5 1 , Vol. 8, No. 5, p. 489 (co-
HILLMAN, A. P . E d i t o r of E l e m e n t a r y P r o b l e m s and Solutions: Vol. 8,

No. 1, pp. 105-112; Vol. 8, No. 3 , pp. 325-334; Vol. 8, No. 4 , pp.
443-448; Vol. 8, No. 5, pp. 542-551.
H O E R B E L T , BERNARD G.
Problem
Solved: B - 1 7 2 , Vol. 8, No. 4 , p. 445;
556
[Dec.
VOLUME INDEX
HOGGATT, VERNER E. , J R . "On Solving CR+2 = CR+1 + CR + n

by E x pansions and O p e r a t o r s / ' V o l . 8, No. 1, pp. 39-48 ( c o - a u t h o r , R. J.
Weinshenk). "Convolution T r i a n g l e s for G e n e r a l i z e d Fibonacci N u m b e r s , " Vol. 8, No. 2 , pp. 158-171. "An Extension of Fibonacci N u m b e r s n , " Vol. 8, No. 2 , pp. 199-216 (co-author, J o s e p h Arkin).
"Some U n i v e r s a l C o u n t e r e x a m p l e s , " Vol. 8, No. 3 , pp. 242-248 (coa u t h o r s , Nanette Cox and John W. P h i l l i p s ) . "An Application of the
L u c a s T r i a n g l e , " Vol. 8, No. 4 , pp. 360-364. " C o m b i n a t o r i a l P r o b l e m s for Generalized Fibonacci N u m b e r s , " Vol. 8, No. 5, pp. 4 5 6 462. P r o b l e m s P r o p o s e d : B - 1 6 2 , Vol. 8, No. 1, p. 108; H - 1 4 1 , Vol.
8, No. 3 , p. 272 (co-proposed by H. T. L e o n a r d , J r . ) ; B - 1 9 3 , Vol. 8,
No. 4 , p. 444. P r o b l e m Solved: B - 1 6 2 , Vol. 8, No. 1, p. 110.
HOMER, JOHN E. P r o b l e m s Solved: B - 1 6 3 , Vol. 8, No. 1, p . 110; B - 1 6 4 ,
Vol. 8, No. 1, p. I l l ; B - 1 7 2 , Vol. 8, No. 4 , p. 445; B - 1 7 3 , Vol. 8,
No. 4 , p. 445; B - 1 7 6 , Vol. 8, No. 4 , p. 448; B - 1 7 7 , Vol. 8, No. 4 ,
p. 448.
HUNTER, J. A. H. " F i b o n a c c i to the R e s c u e , " Vol. 8, No. 4 , p. 406.
P r o b l e m P r o p o s e d : B - 1 6 4 , Vol. 8, No. 1, p. 111. P r o b l e m Solved:
B - 1 6 4 , Vol. 8, No. 1, p. 111.
IVIE, JOHN. P r o b l e m P r o p o s e d : B - 1 6 1 , Vol. 8, No. 1, p. 107. P r o b l e m s Solved: B - 1 6 1 , Vol. 8, No. 1, p . 108; B - 1 6 3 , Vol. 8, No. 1,
p. 110; B - 1 7 2 , Vol. 8, No. 4 , p. 445; B - 1 7 3 , Vol. 8, No. 4 , p. 445.
JAISWAL, D. V. P r o b l e m s Solved: H - 1 4 1 , Vol. 8, No. 3 , p. 275;
Vol. 8, No. 3 , p. 275; H - 1 4 4 , Vol. 8, No. 4 , p . 386; H - 1 4 7 ,
No. 4 , p. 389; H - 1 5 0 , Vol. 8, No. 4 , p. 392; H - 1 5 2 , Vol. 8,
p. 492; H - 1 5 3 , Vol. 8, No. 5, p. 493; H - 1 5 5 , Vol. 8, No. 5,
H - 1 2 6 , H - 1 2 7 , H - 1 2 9 , H - 1 3 1 , Vol. 8, No. 5, p. 498.
H-142;
Vol. 8,
No. 5,
p. 498;
JORDAN, J. H. "A L i m i t e d A r i t h m e t i c on Simple Continued F r a c t i o n s , " Vol.

8, No. 2 , pp. 135-157 ( c o - a u t h o r , C. T. Long).
KARST, EDGAR. "Magic S q u a r e s Consisting of P r i m e s in A. P . , " Vol. 8,
No. 3 , pp. 317-324. "Application of R e c u r s i v e Sequences to Diophantine E q u a t i o n s , " Vol. 8, No. 5 , pp. 463-469 ( c o - a u t h o r s , Raphael
F i n k e l s t e i n and Hymie London).
KING, BRUCE W. P r o b l e m P r o p o s e d : B - 1 8 4 , Vol. 8, No. 3 , p. 325.
P r o b l e m s Solved: B - 1 6 0 , Vol. 8, No. 1, p. 107; B - 1 6 1 , Vol. 8, No.
1, p. 108; B - 1 6 3 , Vol. 8, No. 1, p. 110; B - 1 7 2 , Vol. 8, No. 4 , p.
445; B - 1 7 3 , Vol. 8, No. 4 , p. 445.
KLARNER, DAVID A. P r o b l e m P r o p o s e d : H - 1 6 8 , Vol. 8, No. 1, p. 75.
P r o b l e m s Solved: H - 1 4 9 , Vol. 8, No. 4 , p. 391; H - 1 5 1 , Vol. 8, No.
5, p. 498 (co-solved by M. L. J. Hautus).
KRAVITZ, SIDNEY. " T h e L u c a s - L e h m e r T e s t for M e r s e n n e
Vol, 8, No. 1, pp. 1-3.
Numbers,"
1970]
VOLUME INDEX
557
KUNG, S. H. L. P r o b l e m P r o p o s e d : B - 1 6 8 , Vol. 8, No. 3 , p. 328.

l e m Solved: B - 1 6 8 , Vol. 8, No. 3 , p. 329.
Prob-
LEDIN, GEORGE, J R . P r o b l e m s P r o p o s e d : H - 1 7 3 , Vol. 8, No. 4 , p. 383;

H - 1 4 7 , Vol. 8, No. 4 , p. 387. P r o b l e m Solved: H - 1 4 7 , Vol. 8, No.
4 , p. 389.
LEONARD, H. T. , JR. P r o b l e m P r o p o s e d : H - 1 4 1 , Vol. 8, No. 3 , p. 272
(co-proposed by V. E. Hoggatt, J r . ) .
LEWIS, JOHN B. "The Educational Value in M a t h e m a t i c s / 1 Vol. 8, No. 5,
pp. 522-530.
LIND, DOUGLAS. "A Fibonacci C i r c u l a n t , " Vol. 8, No. 5, pp. 449-455.
P r o b l e m s P r o p o s e d : H - 1 4 0 , Vol. 8, No. 1, p. 81; B-179 (based on
B - 1 6 5 ) , Vol. 8, No. 1, p. 105; B - 1 6 5 , Vol. 8, No. 1, p. I l l ; H - 1 7 1 ,
Vol. 8, No. 3 , p. 268; H - 1 4 5 , Vol. 8, No. 4 , p. 386. P r o b l e m s
Solved: H-140, Vol. 8, No. 1, p. 81; B - 1 6 0 , Vol. 8, No. 1, p. 107;
B - 1 6 1 , Vol. 8, No. 1, p. 107; B - 1 6 3 , Vol. 8, No. 1, p. 110; B - 1 6 5 ,
Vol. 8, No. 1, p. 112; H - 1 4 5 , Vol. 8, No. 4 , p. 386; H-149, Vol. 8,
No. 4 , p. 391.
LINDSTROM, P E T E R A. P r o b l e m s Solved: B - 1 6 3 , Vol. 8, No. 1, p. 110;
Vol. 8, No. 4 , p. 445; B - 1 7 3 , Vol. 8, No. 4 , p. 445; B - 1 7 6 , Vol. 8,
No. 4 , p. 448.
LONDON, HYMIE. "Application of R e c u r s i v e Sequences toDiophantine E q u a t i o n s , " Vol. 8, No. 5, pp. 463-469 ( c o - a u t h o r s , Raphael F i n k e l s t e i n
and E d g a r K a r s t ) .
LONG, C. T. "A L i m i t e d A r i t h m e t i c on Simple Continued F r a c t i o n s , " Vol.
8, No. 2, pp. 135-157 (co-author, J. H. J o r d a n ) .
LU, K. U. " T h e S m a l l e s t N u m b e r with D i v i s o r s a P r o d u c t of
P r i m e s , " Vol. 8, No. 4 , pp. 380-382.
Distinct
LYNCH, W. C. " T h e t - F i b o n a c c i N u m b e r s and Polyphase S o r t i n g , " Vol. 8,

No. 1, pp. 6-22.
LYNN, BRUCE.
P r o b l e m Solved: B - 1 7 2 , Vol. 8, No. 4 , p . 445.
MADACHY, JOSEPH.
pp. 393-396.
E d i t o r , " R e c r e a t i o n a l M a t h e m a t i c s , " Vol. 8, No. 4 ,
MAIER, EUGENE A.
"One-One C o r r e s p o n d e n c e s Between the set
P o s i t i v e I n t e g e r s and the Sets N

365-371.
and
N , " Vol. 8, No. 4 ,
of
pp.
558
[Dec.
VOLUME INDEX
MANA, PHIL. P r o b l e m s P r o p o s e d : B - 1 6 3 , Vol. 8, No. 1, p. 110; B - 1 8 9 ,

Vol. 8, No. 3 , p. 326; B - 1 7 0 , Vol. 8, No. 3 , p . 331; B - 1 7 1 , Vol.
8, No. 3 , p. 333; B - 1 9 4 , Vol. 8, No. 4 , p. 444; B - 1 7 6 , Vol. 8, No.
4 , p. 44?; B - 1 7 7 , Vol. 8, No. 4 , p. 448; B - 1 9 7 , Vol. 8, No. 5 5 p.
542; B - 1 9 8 , Vol. 8, No. 5, p. 543. P r o b l e m s Solved: B - 1 6 3 , Vol.
8, No. 1, p. 110; B - 1 6 7 , Vol. 8, No. 3 , p. 328; B - 1 7 0 , Vol. 8, No.
3 , p. 332; B - 1 7 1 , Vol. 8, No. 3 , p. 337; B - 1 7 6 , Vol. 8, No. 4 , p.
448; B - 1 7 7 , Vol. 8, No. 4 , p. 448.
MILSOM, JOHN W. P r o b l e m s Solved: B - 1 6 0 , Vol. 8, No. 1, p. 107; B - 1 6 8 ,
Vol. 8, No. 3, p. 329; B - 1 7 2 , Vol. 8, No. 4 , p. 445; B - 1 7 3 , Vol. 8,
No. 4 , p. 445; B - 1 7 6 , Vol. 8, No. 4 , p. 448.
MOORE, RICHARD E. M. " M o s a i c Units: P a t t e r n S i z e s in Ancient M o s a i c s / '
Vol. 8, No. 3 , pp. 281-310.
MOST, MEL. P r o b l e m s P r o p o s e d : B - 1 7 4 , Vol. 8, No. 4 , p. 445; B - 2 0 1 ,
Vol. 8, No. 5, p. 543. P r o b l e m Solved: B - 1 7 4 , Vol. 8, No. 4 , p. 446.
ONDERDONK, Philip B.
5, pp. 507-508.
" P i n e a p p l e s and Fibonacci N u m b e r s , " Vol. 8, No.
PADILLA, GLORIA C. P r o b l e m s P r o p o s e d : B - 1 7 2 , Vol. 8, No. 4 , p. 444,

B - 1 7 3 , Vol. 8, No. 4 , p. 445. P r o b l e m s Solved: B - 1 7 2 , Vol. 8, No.
4 , p . 445; B - 1 7 3 , Vol. 8, No. 4 , p. 445.
PARBERRY, EDWARD A. "On P r i m e s and P s e u d o - P r i m e s Related to the
Fibonacci S e q u e n c e , " Vol. 8, No. 1, pp. 49-60.
PARKER, F . D.
P r o b l e m Solved: B - 1 7 0 , Vol. 8, No. 3 , p. 331.
P E C K , C. B. A. P r o b l e m s Solved:
Vol. 8, No. 3, p. 328; B - 1 6 8 ,
No. 3 , p. 332; B - 1 7 1 , Vol. 8,
p. 392; B - 1 7 2 , Vol. 8, No. 4 ,
B - 1 7 4 , Vol. 8, No. 4 , p. 446;
Vol. 8, No. 4, p. 448; B - 1 7 7 ,
B - 1 6 6 , Vol. 8, No. 3 , p. 327;

Vol. 8, No. 3 , p. 329; B - 1 7 0 ,
No. 3 , p. 334; H - 1 5 0 , Vol. 8,
p. 445; B - 1 7 3 , Vol. 8, No. 4 ,
B - 1 7 5 , Vol. 8, No. 4 , p. 447;
Vol. 8, No. 4 , p . 448.
B-167,
Vol. 8,
No. 4 ,
p. 445;
B-176,
P H I L L I P S , JOHN W. "Some U n i v e r s a l C o u n t e r e x a m p l e s , " :Vol. 8, No. 3 ,

pp. 242-248 ( c o - a u t h o r s , Nanette Cox and V. E . Hoggatt, J r . ) .
POLLACK, RICHARD. " R e c u r r e n c e F o r m u l a s , " Vol. 8, No. 1, pp.
(co-author, J o s e p h Arkin).
RAMANNA, J. P r o b l e m P r o p o s e d : H - 1 5 3 , Vol. 8, No. 5, p. 492.
Solved: H - 1 5 3 , Vol. 8, No. 5, p . 492.
4-5
Problem
READ, B . A. " F i b o n a c c i S e r i e s in the Solar S y s t e m , " Vol. 8, No. 4 , pp.

428-438.
R E C K E , KLAUS-GUNTHER. P r o b l e m s Solved: B - 1 7 2 , Vol. 8, No. 4 , p.
445; B - 1 7 3 , Vol. 8, No. 4 , p. 445; B - 1 7 4 , Vol. 8, No. 4, p. 446;
Vol. 8, No. 4 , p. 448.
RENNIE, MICHAEL.
P r o b l e m Solved: B - 1 7 2 , Vol. 8, No. 4 , p. 445.
1970]
VOLUME INDEX
559
ROBERTS, J B. P r o b l e m s P r o p o s e d : B - 1 8 1 , Vol. 8, No. 1, p. 106; B - 1 8 1 ,

Vol. 8, No. 5, p. 549. P r o b l e m Solved: B - 1 8 1 , Vol. 8, No. 5, p. 549.
SATLOW, GERALD. P r o b l e m s Solved:
B - 1 7 2 , Vol. 8, No. 4 , p. 445.
B-167,
Vol. 8, No. 3, p. 328;
SHANNON, A. G. P r o b l e m s P r o p o s e d : B - 1 8 8 , Vol. 8, No. 3 , p. 326; B - 1 6 7 ,

Vol. 8, No. 3 , p. 328. P r o b l e m s Solved: B - 1 6 0 , Vol. 8, No. 1, p.
107; B - 1 6 3 , Vol. 8, No. 1, p. 110; H - 1 4 1 , Vol. 8, No. 3, p. 275;
H - 1 4 3 , Vol. 8, No. 3 , p. 379; B - 1 6 7 , Vol. 8, No. 3 , p. 328; H - 1 5 0 ,
Vol. 8, No. 4, p. 392; B - 1 7 2 , Vol. 8, No. 4 , p. 445; B - 1 7 3 , Vol. 8,
No. 4 , p. 445; B - 1 7 4 , Vol. 8, No. 4, p. 446; B - 1 7 5 , Vol. 8, No. 4 ,
p. 447; B - 1 7 6 , Vol. 8, No. 4 5 p. 448.
S I E L A F F , RICHARD W.
P r o b l e m Solved: B - 1 7 2 , Vol. 8, No. 4 S p. 445.
SIMMONS, GUSTAVUS J. P r o b l e m s P r o p o s e d : B - 1 8 3 , Vol. 8, No. 1, p. 106;

B - 1 8 3 , Vol. 8, No, 5, p. 551. C o m m e n t : B - 1 8 3 , Vol. 8, No. 5, p.
551.
SINGMASTER, DAVE). "Some C o u n t e r e x a m p l e s and P r o b l e m s on
R e c u r r e n c e R e l a t i o n s / 1 Vol. 8, No. 3 , pp. 264-267.
Linear
SITGREAVES, ROSEDITH. "Some P r o p e r t i e s of Stirling N u m b e r s of the

Second K i n d , " Vol. 8, No. 2, pp. 172-181.
SPICKERMAN, W. R.
397-401.
"A Note on Fibonacci F u n c t i o n s , " Vol. 8, No. 4 , pp.
STOLARSKY, KENNETH B. "Infinitely Many G e n e r a l i z a t i o n s of A b e l ' s P a r tial Summation I d e n t i t y , " Vol. 8, No. 4 , pp. 375-379.
SURYANARAYANA, D.
"A T h e o r e m Concerning Odd P e r f e c t N u m b e r s , "
Vol. 8, No. 4, pp. 337-346 ( c o - a u t h o r , P e t e r H a g i s , J r . ) .
SWAMY, M. N. S. P r o b l e m s P r o p o s e d : H - 1 5 0 , Vol. 8, No. 4 , p. 391;
H - 1 5 5 , Vol. 8, No. 5 , p. 497. P r o b l e m s Solved: H - 1 5 0 , Vol. 8, No.
4 , p. 391; H - 1 5 5 , Vol. 8, No. 5, p. 497.
TRIGG, CHARLES W. P r o b l e m s Solved: B - 1 7 2 , Vol. 8, No. 4, p. 445;
Vol. 8, No. 4, p. 448; B - 1 7 7 , Vol. 8, No. 4 , p. 448.
UMANSKY, HARLAN L. " L e t t e r s to the E d i t o r , " Vol. 8, No. 1, pp. 88-89
( c o - a u t h o r , David E. F e r g u s o n ) .
WALL, CHARLES R. P r o b l e m s P r o p o s e d : H - 1 4 3 , Vol. 8, No. 3, p. 277;
H - 1 4 9 , Vol. 8, No. 4 , p 0 390. P r o b l e m s Solved: H - 1 4 3 , Vol. 8, No.
3 , p. 279; H - 1 4 9 , Vol. 8, No. 3 , p. 390.
m
WEINSHENK, R. J. "On Solving C _ ) _ = C , 1 + C
+n
b y E x p a n s i o n s and
O p e r a t o r s , " Vol. 8, No. 1, pp. 39-48 (co-author, V. E. Hoggatt, J r . ) .
WESSNER, JOHN. P r o b l e m s Solved: B - 1 6 7 , Vol. 8, No. 3 , p . 328; B - 1 7 0 ,
Vol. 8, No. 3 9 p. 332; B - 1 7 1 , Vol. 8, No. 3 , p . 334; B - 1 7 2 , Vol. 8,
No. 4 , p. 445; B - 1 7 4 , Vol. 8, No. 4 , p 9 446.
560
Dec. 1970
VOLUME INDEX
WHITNEY, RAYMOND E. E d i t o r of "Advanced P r o b l e m s and Solutions,"

Vol. 8, No. 1, p p . ' 7 4 - 8 2 ; Vol. 8, No. 3 , pp. 268-279; Vol. 8, No.
4 , pp. 383-392; Vol. 8, No. 5, pp. 487-498. "On a C l a s s of Difference
E q u a t i o n s , " Vol. 8, No. 5, pp. 470-475.
WILLIAMS, J . M. , J R .
515.
" T h e P o w e r s of T h r e e , " Vol. 8, No. 5, pp. 509-
WULCYZN, GREGORY. P r o b l e m s Solved: B - 1 6 0 , Vol. 8,

B - 1 6 1 , Vol. 8, No. 1, p . 108; B - 1 6 4 , Vol. 8, No. 1,
Vol. 8, No. 4 , p. 445; B - 1 7 3 , Vol. 8, No. 4 , p . 445;
No. 4 , p . 447; B - 1 7 6 , Vol. 8, No. 4, p . 448; B - 1 7 7 ,
p. 448.
No. 1,
p. I l l ;
B-175,
Vol. 8,
p. 107;
B-172,
VoL 8,
No. 4 ,
YALAVIGI, C. C. P r o b l e m s P r o p o s e d : B - 1 6 9 , Vol. 8, No. 3 , p. 329;

H-176, Vol. 8, No,, 5, p. 448. P r o b l e m s Solved: B - 1 6 9 , Vol. 8, No.
3 , p . 331.
YAP, H. P . "A Simple R e c u r r e n c e Relation on Finite Abelian
VoL 8, No. 3 , pp. 255-263.
Groups,"
YODER, MICHAEL. P r o b l e m s Solved: H - 1 3 7 , Vol. 8, No. 1, p. 76;

Vol. 8, No. 1, p. 82; B - 1 6 0 , Vol. 8, No. 1, p. 107; B - 1 6 1 ,
No. 1, p . 108; B - 1 6 2 , VoL 8, No. 1, p. 109; B - 1 6 3 , VoL 8,
p . 110; B - 1 6 4 , Vol. 8, No. 1, p . I l l ; B - 1 6 5 , VoL 8, No. 1,
H - 1 4 3 , VoL 8, No, 3 , p . 279; H - 1 4 4 , VoL 8, No. 4 , p . 386;
VoL 8, No. 4, p. 387; H - 1 4 7 , VoL 8, No. 3 , p. 389; H - 1 4 9 ,
No. 3 , p. 391; H - 1 5 0 , Vol. 8, No. 3 , p. 392; B - 1 7 2 , Vol. 8,
p. 445; B - 1 7 3 , Vol. 8, No. 4 , p. 445; B - 1 7 4 , Vol. 8, No. 1,
B - 1 7 5 , VoL 8, No. 4 , p. 446; B - 1 7 6 , VoL 8, No. 4 , p. 448;
VoL 8, No. 4 , p . 448.
H-140,
Vol. 8,
No. 1,
p . 112;
H-145,
Vol. 8,
No. 4 ,
p. 446;
B-177,
ZAREMBA, S. K. "A R e m a r k a b l e L a t t i c e G e n e r a t e d by Fibonacci N u m b e r s , "

VoL 8, No. 2, pp. 185-198.
ZEITLIN, DAVID. "On D e t e r m i n a n t s Whose E l e m e n t s a r e P r o d u c t s of R e c u r s i v e S e q u e n c e s , " Vol. 8, No. 4 , pp. 350-359. P r o b l e m s P r o p o s e d :
B-166. (suggested by solutions to B - 1 4 8 , B - 1 4 9 , and B-150), VoL 8,
No. 3 , p. 326; B - 1 9 5 , Vol. 8, No. 4, p. 444; B-175 (composed from
solution to B - 1 5 5 ) , Vol. 8, No. 4 , p. 446. P r o b l e m s Solved: B - 1 6 3 ,
Vol. 8, No. 1, p . 110; H - 1 4 1 , VoL 8, No. 3 , p . 273; H - 1 4 3 , VoL 8,
No. 3 , p. 277; B - 1 6 6 , VoL 8, No. 3, p. 327; B - 1 7 0 , Vol. 8, No. 3,
p. 332; B - 1 7 1 , Vol. 8, No. 3 , p . 334; H - 1 4 7 , Vol. 8, No. 4 , p . 388;
B - 1 7 3 , Vol. 8, No. 4 , p . 445; B - 1 7 4 , VoL 8, No. 4 , p . ' 4 4 6 ; B - 1 7 5 ,
Vol. 8, No. 4 , p. 447; B - 1 7 6 , VoL 8, No. 4 , p . 447; B - 1 7 7 , VoL 8,
No. 4 , p. 448; H - 1 5 5 , VoL 8, No. 5, p. 498.
SUSTAINING MEMBERS
*H, L. A l d e r
V. V. A l d e r m a n
G. L. A l e x a n d e r s o n
R. H. Anglin
* Joseph Arkin
L. L. Badii
Don B a k e r
Col. R. S. B e a r d
B. C. B e l l a m y
Murray Berg
Leon B e r n s t e i n
* M a r j o r i e Bicknell
J . H. Biggs
J. J. Bohnert
M. B. B o i s e n , J r .
T. A. B r e n n a n
C A. B r i d g e r
Maxey Brooke
* B r o . A. B r o u s s e a u
* J . L. B r o w n , J r .
C. R. Burton
P. F. Byrd
N. S. C a m e r o n
L. C a r l i t z
P . B. C h a r l a n d
P. J. Cocussa
L e e Corbin
Nannette Cox
A. B. C u m m i n g s
D. E. Daykin
J . W. D e c e l i s
F r a n c i s DeKoven
J . E . Desmond
A. W. Dickinson
M. H. Diem
N. A. D r a i m
* Charter Members
D. C. Duncan
M. H. E a s t m a n
Merritt Elmore
Re S. E r l i e n
H. W. E v e s
R. A. F a i r b a i r n
A. J. F a u l c o n b r i d g e
H. H. F e r n s
D. C. F i e l d e r
E. T. F r a n k e l
C. L. G a r d n e r
G. H. Glabe
*H. W. Gould
Nicholas Grant
G. B. G r e e n e
G. A. Guillotte
B. H. Gundlach
V. C. H a r r i s
W. R. H a r r i s , J r .
*A. P . Hillman
M r . and M r s . B. H. H o e l t e r
*V. E. Hoggatt, J r .
*A. F . H o r a d a m
J . A . H . Hunter
D. F . Howells
F . R. J a c h i m
A. S. J a c k s o n
*Dov J a r d e n
J . H. J o r d a n
R. P . K e l i s k y
* Kenneth K l o s s
D. E. Knuth
* Sidney K r a v i t z
George Ledin, J r .
P . F. Lehman
Eugene Levine
* D . A. Lind
* C . T. Long
A. F . Lopez
F . W. Ludecke
J . S. Madachy
* J . A. Maxwell
* S i s t e r M a r y deSales McNabb
W. A. Morin
L u c i l e Morton
Stephen Nytch
R. Do 0 ? C o n n e l l
P . B. Onderdonk
F . J. O s s i a n d e r
Ann P a p e
R. J. P e g i s
J . W. P h i l l i p s
Alwin P o r t n e y
Mo M. Risueno
* D . W. Robinson
F . G. Rothwell
A. M. Routh
B. B. Sharpe
L. R. Shenton
J . A. Shumaker
D. C. Stone
M. N. S. Swamy
A. Sylwester
* D . E. T h o r o
H. L. Umansky
M. E. Waddill
* L . A. Walker
R. J . Weinshenk
R. A. White
R. E. Whitney
P . A. Willis
C h a r l e s Ziegenfus
ACADEMIC OR INSTITUTIONAL MEMBERS

DUKE UNIVERSITY
D u r h a m , No. C a r o l i n a
ST. MARY'S COLLEGE

St. M a r y 7 s C o l l e g e , California
NORWICH UNIVERSITY
Northfield, V e r m o n t
VALLEJO UNIFIED SCHOOL DISTRICT

V a l l e j o , California
SACRAMENTO STATE COLLEGE

S a c r a m e n t o , California
WAKE FOREST UNIVERSITY

W i n s t o n - S a l e m , North C a r o l i n a
SAN JOSE STATE COLLEGE

San J o s e , California
WASHINGTON STATE UNIVERSITY

P u l l m a n , Washington
SONOMA STATE COLLEGE

R o h n e r t P a r k , California
THE CALIFORNIA MATHEMATICS

COUNCIL
BINDERS NOW AVAILABLE

The F i b o n a c c i A s s o c i a t i o n is making available a binder which
can be used to take c a r e of one volume of the p u b l i c a t i o n at a t i m e .
This binder is d e s c r i b e d as follows by the company p r o d u c i n g it:
" . . . . The binder is made of heavy weight v i r g i n vinyl,
e l e c t r o n i c a l l y sealed over r i g i d b o a r d equipped with
a c l e a r label holder extending 2 - 3 / 4 " high from the
bottom of the backbone, round c o r n e r e d , fitted w i t h
a 1 1/2 " multiple m e c h a n i s m and 4 heavy w i r e s . "
The n a m e , FIBONACCI QUARTERLY, is p r i n t e d in gold on the
front of the binder and the s p i n e . The color of the binder is d a r k
green.
T h e r e is a s m a l l pocket on the spine for holding a tab
giving y e a r and v o l u m e . These l a t t e r w i l l be supplied with e a c h
o r d e r if the volume or volumes to be bound a r e indicated.
The p r i c e p e r binder is $ 3 . 5 0 which includes postage ( r a n g i n g
f r o m 50 to 80 for one b i n d e r ) . The tabs will be s e n t w i t h the
r e c e i p t or invoice.
All o r d e r s should be sent to: B r o t h e r Alfred B r o u s s e a u ,
Managing E d i t o r , St. Mary* s College, Calif. 94575

The Fibonacci Quarterly: Number

Uploaded by

Document Informationclick to expand document informationThe Fibonacci Quarterly

Document Informationclick to expand document information

Copyright:

Available Formats

The Fibonacci Quarterly: Number

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Fibonacci Quarterly: Number

Uploaded by

Copyright:

Available Formats

r^

THE FIBONACCI QUARTERLY

THE OFFICIAL JOURNAL

Verner E. Hoggatt, jr. 456

Raphael Finkelstein, Edgar Karst, and Hymie London 463

On a Class of Difference Equations

RaymondE. Whitney 470

David G. Beverage 498

Determination of Heronian Triangles

John H. Carlson 499

Pineapples and Fibonacci Numbers . . . . . . .

Philip B. Onderdonk 507

The Powers of Three

J.M. Williams, Jr. 509

Some Remarks on the Ordering

Joe R. Crenshaw 516

. Joseph Arkin 531

Fibonacci Numbers as Paths

Edward T. Frankel 538

Elementary Problems and Solutions

Edited by A . P. Hillman 542

THE FIBONACCI QUARTERLY

All subscription correspondence should be addressed to Brother Alfred

denotes the Fibonacci numbers defined by

bytheVandermonde determinant V = Jco?I (i s j = 0S 1, 8 8 , n - 1). Denoting

" " > - 0 ( 1 - 0 k ) ( l - /3w k )

r " F n + r " < F n+r-l " F r - l ) W k

Now for any x and y ,

CFr " F n + r " ( F n + r - l ~

" <Fn+r-l " V l * "

n a - aoj ){i - pw ) n a - aoj ) n (i = (1 - an)(l

simplifies if n i s even* Ruggles

2. The generalization of (1) to second-order recurring sequences uses

WQ and Wj arbitrary, where p 2 - 4q ^ 0. Let a and b be the roots of

root of unity* Since the roots are distinct, there

are constants A and B such that W

Define the sequence

which agrees with (1) by taking p = 1, q = - 1 , W = F , and V

3, We now consider a slight variant of the above.

A determinant of the form

is termed a skew circulanfc,

Scott [1, Vol. 4, p. 356] has shown that

(2k + l)ir __, . . (2k + l)ir

r o o t s of - 1 . To p r o v e (6) quickly, multiply S(a 0 , , a

the Vandermonde d e t e r m i n a n t | e] | (i, j = 0S 1, * , n - 1),

the proof of (3).

Application of this to (7) yields the d e s i r e d r e s u l t (5).

Then if n s 0 (mod 4),

while if n = 2 (mod 4),

The extension of this to s e c o n d - o r d e r r e c u r r i n g s e q u e n c e s involves

no new i d e a s , and the d e t a i l s a r e t h e r e f o r e omitted.

Using (6) and (8), we find

T h o m a s M u i r , The T h e o r y of D e t e r m i n a n t s (4 V o l s . ) , D o v e r , New York,

2* L D. R u g g l e s , "Some Fibonacci R e s u l t s Using F i b o n a c c i - T y p e S e q u e n c e s / '

V* I. S m i r n o v , L i n e a r A l g e b r a and Group T h e o r y , M c G r a w - H i l l , New

Theorem 1. The number of subsets of ( l , 2, 3, , n} which have

] is the binomial coefficient.