2 1

THE FIBONACCI QUARTERLY
THE OFFICIAL JOURNAL OF

VOLUME 2
THE FIBONACCI ASSOCIATION
NUMBER 1
PART 1 ADVANCED
A Property of Fibonacci Numbers

On Square Lucas Numbers . . . . . . . . . .
Lattice Paths and Fibonacci Numbers
A Note on Fibonacci Numbers
R. L. GRAHAM
BROTHER
ROBERT
U. ALFRED
E. GREENWOOD
L. CARLITZ
1
11
13
15
Some New Fibonacci Identities

.
. VERNER E. HOGGATT, JR. and MARJORIE BICKNELL
29
Primes Which Are Factors Of All Fibonacci Sequences . . BROTHER U. ALFRED

A Fibonacci Test For Convergence
j . H. JORDAN
Exploring Fibonacci Residues
. BROTHER U. ALFRED
Transcendental Numbers Based On The Fibonacci Sequence . . . DONALD KNUTH
Strengthened Inequalities For Fibonacci And Lucas Numbers . . . . DOV JARDEN
Advanced Problems And Solutions . . . . Edited by VERNER E. HOGGATT, JR.
33
39
42
43
45
49
PART 1 1 ELEMENTARY
Beginners' Corner . . . . . . . . . . .
A New Proof For An Old Property
Edited by DMITRI
THORO
GLEN MICHAEL
A Primer For The Fibonacci Numbers - Part V

. . . , VERNER E. HOGGATT, JR. and I. D. RUGGLES
On The General Term Of A Recursive Sequence

Elementary Problems And Solutions
FEBRUARY
. . . . . FRANCIS
Edited by A .
D. PARKER
P. HILLMAN
53
57
59
67
72
1964
THE FIBONACCI QUARTERLY

OFFICIAL ORGAN OF
THE FIBONACCI ASSOCIATION

A JOURNAL DEVOTED TO THE STUDY OF INTEGERS WITH SPECIAL PROPERTIES
EDITORIAL BOARD
Brother U. Alfred
H. L. Alder
S. L. Basin
John L. Brown, J r .
H. W. Gould
A. P. Hillman
V. E. Hoggatt, J r .
Leo Moser
I. D. Ruggles
D. E. Thoro
WITH THE COOPERATION OF
P. M. Anselone
Charles H. King
Terry Brennan
L. H. Lange
Maxey Brooke
Douglas Lind
Paul F. Byrd
Calvin T. Long
Leonard Carlitz
James Maxwell
Calvin D. Crabill
Sister M. de Sales McNabb
H. W. Eves
C. D. Olds
John H. Halton
D. W. Robinson
Richard A. Hayes
Azriel Rosenfeld
A. F. Horadam
John E. Vinson
Dov Jar den
Lloyd Walker
Stephen Jerbic
Charles R. Wall
R. P. Kelisky
The California Mathematics Council, Northern Section
and
The San Jose State College Foundation
All subscription correspondence should be addressed to Brother U. Alfred, St.
MaryTs College, Calif* All checks ($4.00 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 Jose, Calif. 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 ! s College Post
Office, California, as an official publication of the Fibonacci Association.
A PftOPEBTY OF FIBONACCI LUMBERS

R. L. GRAHAM
Bell Telephone Laboratories, Inc., Murray H i l l , N. J.
1.
INTRODUCTION
L e t A - (a l s ar, s - ) denote a (possibly finite) sequence of i n t e g e r s .
We
oo
shall l e t P(A) denote the s e t of all i n t e g e r s of the f o r m
2 e, a, w h e r e e,
K
k=l K K
i s 0 o r 1. If all sufficiently l a r g e i n t e g e r s belong to P(A) then A i s s a i d
^ ^ e c o m p l e t e . F o r e x a m p l e , if F = ( F l s F 2 , - ) , w h e r e F i s the n
Fibonacci n u m b e r , i. e. , F n = 0, F1 = 1 and F
= F , _, + F for n ^ 0. then
u
J
1
?
'
n+2
n+1
n
F i s complete (cf. [1] ). M o r e generally 9 it can be e a s i l y shown that F s a t i s fies the following conditions:
(A) If any one t e r m i s r e m o v e d from
F then the r e s u l t i n g sequence
is
complete.
(B) If any two t e r m s a r e r e m o v e d from F then the r e s u l t i n g sequence i s
not c o m p l e t e .
(A s i m p l e proof of (A) i s given in [l] ; (B) will be p r o v e d in Section 2 . )
In this p a p e r i t will be shown that a " s l i g h t " modification of F
a r a t h e r s t a r t l i n g change in the additive p r o p e r t i e s of F .
sequence S which h a s F
- (-1)
as its n
produces
In p a r t i c u l a r ,
the
t e r m has the following r e m a r k a b l e
properties:
(C) If any finite subsequence i s deleted from S then the r e s u l t i n g sequence
is complete.
(D) If any infinite subsequence i s deleted from
S then the r e s u l t i n g s e -
quence i s not c o m p l e t e .
2.
THE MAIN RESULTS
We f i r s t prove (B). Suppose F

F*
(where r < s ) .
0,1,2, .
and F
a r e r e m o v e d from
We show by induction that F
^ i
"
F to form
$ P(F*) for
We f i r s t note that the s u m of all t e r m s of F* which do not exceed
F s + 1 - 1 is just
^\
k=l
k =
= }] < F k + 2 " F k + 1> k=l
s+1 "
~ Fr *
s+1 "
A PROPERTY OF FIBONACCI NUMBERS
and hence F
- 1 ^ P(F*) Now assume that F
t ^ 0 and consider the integer F , ? . - 1.

which are less than F
+2t+9
is
[Feb.
- 1 <j: P(F*) for some
The sum of all terms of F*
118
J ^
s+2t+l
F, ~ F - F = F
- 1 - F - F < F
k
r
s
s+2t+3 x
r
s
s+2t+3
-1
x
k=l
Thus, in order to have F , ? t + o - 1 e P(F*) we must have F + t + o - 1 =
F
s+2t+2 + m ' w h e r e m e P"F*)" B u t m = Fs+2t+3 " Fs+2t+2 " ^ = F s + 2 t + l

- 1 which does not belong to P(F*) by assumption. Hence F 9 , Q - 1 ^
P(F*) and proof of (B) is completed.
We now proceed to the main result of the paper.
Theorem: Let S = (sl9s2f--a)
s
= F - (-l) .
be the sequence of integers defined by
Then S satisfies (C) and (D).
Proof: The proof of (D) will be given first. Let the infinite subsequence
s.l < s. < s. < . . . be deleted from S and denote the remaining sequence by
i
h
h
S*. In order to prove (D) it suffices to show that
3.i
s.
- 1 <f P(S*) for n > 4
We first note that

s.
it
+ s.
i2
>
Sj + s 2 = 2
Therefore, we have (cf. Eq. (1) )

i -1
.,
s. < s. ., - s.
J
1
- s.
**
< s. , - - 2
1
3=1
Hence, to represent s^ + j_ - 1 in P(S*) we must use some term of S* which
exceeds s^ _i (since by above, the sum of all terms of S* not exceeding
S| - i is less than Si +]_ - 1 for n > 4). Since Si
the smallest term of S* which exceeds s.
., is s. , 1 (whichfl of course, is
in-l
greater than s.
t1
1 +1
-.1). Thus
is missingfrom S*5 then

in+l
1964 ]
A P R O P E R T Y OF FIBONACCI NUMBERS
s.
- 1 f P(S*) for n ^
+1
n
and (D) i s proved,,
To p r o v e (C) f
s
F o r non
k+2* * * ^
and l e t S f
let k > 4
*"
ne
ative
denote the sequence
( s , , s,
i n t e g e r s w and x, P(S ) i s s a i d to have no gaps
of length g r e a t e r than w beyond x provided t h e r e do not exist w + 1 c o n s e c u tive i n t e g e r s exceeding x which do not belong to P(S f ).
The proof of (C)
is
now a consequence of the following two l e m m a s .

T h e r e e x i s t s v such that P(S f ) has no gaps of length g r e a t e r
L e m m a 1:
than v beyond s, .
L e m m a 2:
beyond s,
If
w > 0 and P(S ! ) h a s no gaps of length g r e a t e r than w
then t h e r e e x i s t s i such that P(S f ) has no gaps of length g r e a t e r
than w - 1 beyond s. .
Indeed, by L e m m a 1 and r e p e a t e d application of L e m m a 2 i t follows that
t h e r e e x i s t s j such that P(S ! ) has no gaps of length g r e a t e r than 0 beyond
s. .
That i s ,
S f i s c o m p l e t e , which p r o v e s
(C)8
Proof of L e m m a 1: F i r s t note that
*Wi
= F9 - ( - D 2 n + F , ^
- ( - l ) 2 n + 1 = F_
+ F.
+ 1
= F5
S
Similarly,
S
2n+1
2n+2
=
=
_,
, ' 2n+l , _
+ F
2n+1 "
2n+2 "
"P
2n+l
= F
+ F
2n+2
1v 2n+2
= s
2n+3
- 1
2n+3
A l s o , we have
S l + S 2 +
..
S n = ( F l + 1) + ( F 2 - l ) + -- + ( F n - ( - l ) n )
=
F . + en
> (F. +2^ - F . +1n ) + n
j=i
(1)
= F
S
where
l0
n+2
n+2
3=1
- 1 +
2n+2

0
n
Thus
[Feb,
for n even
1 1 for n odd
ii
i-i-
s n+2
, o - sm+1
M - e + e
n m - 1.. for n > m
Now. let h > k + 1 and let

P' = P ( ( s k , s k + l f . - - f s h ) )
={pi,P2,---,P^}
where pj < pj < < p.1 . Let
max
, .
,x
= b r a i - l (pi+l " P P '
Then
h > k + 1 > 5=*sh s sk+1 + 2
-^Sh M
k+i
+ e
h " Vi
*' s h+2 " V l ~ s k+l
+ x
+ e
h " e k+l
h
* s h+l ~ s h+2 " s k+l " e h
+ e
k+l
Since
iSi
k+i
vi>-<p;
+s
h+i>) = v
then in
" = P(<V-'VW)
= p((skJ...,Sh))W{q+sh+1:q
= {Pi,.pi,.-",P^} .
P \^%...
,sh))}
1964]
where p " < pT2T< < p" f , we have
Similarly, since
h+1
h > k + 1 > 5=^>s U l 0 ^ y s.

then in
p- = p((v.v2>) ={p;Mpr'-"'Pn'->
where p m < p m < < pTt
i
n?! *
we have
max
(p"
- p"Jr ) < v9 etc.
r+1
l<r<nn~l
By continuing in this way9 Lemma 1 is proved.
The proof of Lemma 2 is a consequence of the following two results:
(a) For any r ^ 0 there exists t such that m > t implies all the
integers
s m + y9
y = 0 S 1 9 2 , - - - , ( r - 1)
belong to P(S f ).
(b) There exists r f such that for all sufficiently large h ! ? P(Sf) has
no gaps of length greater than w - 1 between s, , + r f and s, f+ - r ? (L e. , there do not exist w consecutive integers exceeding s, ?
+ r ! and less than s, , - - r f which are missing from P(S!) ).
Therefore, for s. sufficiently large f P(Sf) has no gaps of length greater
than w - 1 beyond s., which proves Lemma 2
Proof of (a): Choose p such that
2p - 3 ^
and
2D-2 ~
and choose n such that

n ^ s 2 p. 2
and
n > r + k
Then
n
2n-2m~4
2i-1
i=n-m
n-m-1
S
r l 2i-1 " 1
j=2p-3
i=l
2i-1
2n2m-4
S
i=l
n
n+ F
I 2i-l"
j=2p~3
n-ml
F
n+m+1
"
i=l
1=1
. = m + 1 + F2n - F 2 n _ 2 m ^
= s 2 n - (s 2
[Feb,
2i-1
2n-2m~4
S
j=2p-3
s 2 ^ 2 m _ 2 + 0 - s2p_2- 0
- m - 1), for 0 ^ m < n - p - 1
Since 2p - 3 ^ k, then all the summands used on the left-hand side are in
S\
Hence, all the integers

S
2n -
(s
2p-2 "
"
belong to P(S f ). Since n ^ s 2
X)
>
"
+ pf
n-p - 1 >
"
"" X
then
s2p_2 - 1
Therefore, all the integers

S
2n"
(s
2p-2 "
"
).
2p-2-1
belong to P(S f ), i. e. 9 all the integers

s
But s 2
~ ^ r,
2 n
-m',
os
m's
s2p_2-!..
so that we finally see that all the integers

s
2n ~
m?
'
mf
'~
"
belong to P(S f ).
To obtain sums which exceed s 2 , note that for 1 < m ^ n - k we have
1964]
II
11-111.
n-m.
II
2,
2j-1
+ S
2n-2m
/ , S 2j-1 "
j=n-m+l
j=l
2j-1
2n~2m
j=l
= n + F 0 - (n - m); - F 0 0
+ s0 0
2n
.
2n-2m
2n-2m
= m + F0 - 1
2xi
= m + s~
2n
Since the sums

n
2M
2n-2m
for
2"-n -
j=n~m+l
are all elements of P(Sf),
and since n - k > r,

s 2 n + m
then all the integers
1< m < r
belong to P(S?).
Arguments almost identical to this show that for all sufficiently large n,
all the integers
belong to P(Sf).
2n+1
+ m
>V '*(*- 1)
This proves (a).
Proof of (b): We first give a definition.

finite sequence of integers.
n
the number y
,'
a,.
Let A = (a l9 a 2g -** 9 a ) be a
The point of symmetry of P(A) is defined to be
The reason for this terminology arises from the fact
k=l
that if P(A) is consideredas a subset of the real line^ then" P(A) is symmetric
n
about the point = /
a, .
For we have
k=l
n
P =
n
a
kk
P(A)^> ^ (1 - ^k)ak = ak - p P(A)
k=l
k=l
k=l
n
and the points p and \
P4
n
a ^ - . p are certainly equidistant from i \
ak.
A P R O P E R T Y O F FIBONACCI NUMBERS
[Feb 0
Now note t h a t if r I s sufficiently l a r g e then

s
.
r-1
> 3 > -s, , - + 3 .

k+1
'
s ,i - s
r+1
r
+ 1 - S, , -
r
S
> - s , _,. + 2 9
k+1
'
r+2 "
<
S ,- - 1
k+1
s
r+l "
r+1
<
k+l
'
r+l
r ~
k+l
and
r+2
k+1
r+1
k+1
Therefore
r
2 L S"i =~ 2 ' ( s r + 2 " V l
2 j=k
Z j
2
"
to-l)
<
r+1
and
2(8r+2- V l for all sufficiently l a r g e r .
Vl>
>
In o t h e r w o r d s f for all sufficiently l a r g e r 9
point of s y m m e t r y of P ( ( s , s , s )) l i e s between s,
and s
the
- . By hypothe-
s i s no gaps of length g r e a t e r than w o c c u r in P(S ? ) beyond s , .
Since h > k
> 4 Implies
h+1
h+2
then no gaps of length g r e a t e r than w can o c c u r in P ( ( s ] ? * e , s

K.
s,
and s
< sr+2
...
<
)) between
r
(For if they did, then they would remain in P(S') since s

.)
But
Vl
> 2
j=k
T_
1 )
and
/^
s. i s the point of s y m m e t r y of P((s, , , s )).

j
K
JL
j=k
s. - s L i <
j
r+1
20
j=k
Therefore,
1964]
and by symmetry no gaps of length greater than w occur in P((s, , - , s ))

between
> s. - sr +,1 - and 7 s. - s,
A 3
Li 3
h
Thus, no gaps of length greater than w occur between s,
h "
r+2 " S k+1 ~ h
and
k+l " S h
j=k
provided that r is sufficiently large, Now consider P((s, f , s
)). Since
r+1
r+1
then s
- +s
r+2
r+3
+ p and s
'
+ p are elements of P((s, , . . , s
))
which
differ by 1 whenever . p is an element of P ( ( s , s ? s )) Hence, since in

K
r
P((s k 9 , s )) there are no gaps of length greater than w between s, and
r
? s. - s, , then in P((s k , , s . )) there are no gaps of greater length than
j=k
w - 1 between
sh + s r + 3
Similarly, consider P((s^ s # s s

s
r+2
I.
a*d
r+3
s - sh + s r + 3 .
* ))8 Since
=
r+4
and there are no gaps in P((s k , s

r+1
)) of length greater than w between
s, and V s . - s,s then there are no gaps in P((s, , 9 s

j=k
greater than w - 1 between
s
r.f i
+ B
i+4
and
j=k
j " S h + S r+4
.)) of length
10
A P R O P E R T Y O F FIBONACCI NUMBERS
In g e n e r a l , for q > 0 s i n c e S.
.+ s r + q + 1 = s r + q + 2 + (-1)
[Feb. 1964]
H
and t h e r e a r e
no gaps in P ( ( s k , - . . , s
_ 1 )) of length g r e a t e r than w between s ^ and
r+q-1
2
s. - s U J then t h e r e a r e no gaps in P((s 1 , - - - , s
9 )) of length g r e a t e r
h
T+q+Z
j=k J
r+q-1 k
s
than w - 1 between s h + s r + q + 2 and S
j ~ s h + Vq+2' B u t
r+q-1
S
j "
r+q-2 '"
j:=k
r+q+l
k+1
Vq+1
r+q+1
- s
+ r-n
~ S r+q+3 + ( 1 }
S
r+q+3 " S k+1
'"*k+l
- s
S
k+1
- s
S
h
r+G+2
- e
r+q+1
+ e
k+1
T h e r e f o r e , if we l e t
"
k+1
then for all sufficiently l a r g e z, t h e r e a r e no gaps in P((s, , , s z ) ) of length

g r e a t e r than w - 1
between s + r f and s
- r r (since the p r e c e d i n g a r g u z
z+x
ment i s valid for q > 0 and all sufficiently l a r g e r ) . This c o m p l e t e s the proof
of (b) and the t h e o r e m .
3.
CONCLUDING REMARKS
E x a m p l e s of s e q u e n c e s of positive i n t e g e r s which satisfy both

a r e rather elusive.
(C)and(D)
It would be i n t e r e s t i n g to know if t h e r e e x i s t s such a
quence, say T = ( t i , t 2 , ' - ' )
which i s e s s e n t i a l l y different from S^
see.g.,
such that
l i m *n+l
noo t
n
, 1 + A/5
2
The author w i s h e s to e x p r e s s Ms gratitude to the r e f e r e e for s e v e r a l s u g gestions which made the p a p e r c o n s i d e r a b l y m o r e r e a d a b l e .

REFERENCE
1.
J. L. Brown, "On Complete Sequences of I n t e g e r s , " A m e r . Math. Monthly,

68 (1961) pp. 5 5 7 - 5 6 0 .
m SQUARE LUCAS NUMBERS

B R O T H E R U.
ALFRED
St. M a r y * s C o l l e g e ,
California
Among the first dozen members of the Lucas sequence (1,3,4,7, 11, 18,
) there are two squares, Lt = 1 and L3 = 4. Are there any other square
in the Lucas sequence?
Since the period of the Lucas sequence modulo 8 is 12, it follows that
L.19,
= L (mod 8), so that all possible residues are represented in the fol-
lowing table.
X
12k+X(md8)
0
2
1
1
2
3
3
4
4
7
5
3
6
2
7
5
8
7
9
4
10
3
11
7
It follows that the only Lucas numbers which may be squares are L ' 2 ,
with
X = 1,3 or 9, since the other residues modulo 8 are quadratic non-residues

of 8.
From the general relation
2L
5F F +L
a+b =
z b zH
it follows if t = 2 , r ^ 1, that
2L
5F
X,2t =
5 F
XF2t+LXL2t
AFtLt+LX(Lt-2>
so that
2L
X + 2t
"2LX
(m dL
-L^mod^)
t>
But (L t ,2) = 1. Hence

L
X + 2t
We can use this relation to advantage by writing

11
12
ON SQUARE LUCAS NUMBERS

L
aS
12k+X
[Feb. 1964]
X+2mt
r
w h e r e m i s odd and t = 2 , r 1.
Then
L
X+2mt "~ L X + 2(m-l)t

S+L
(m
( m o d
X + 2(m-2)t
(~l)m K
d V
(mod L,)
X
For
X = 1,
L
12k+1
" L l " ~1
( m
d L
t)?
But
Y) = - 1 , s i n c e
T h e r e f o r e L.. ~,
ilarly,
L-?,
except for
^ 3 (mod 4)
may not be a p e r f e c t s q u a r e except for L^ = !
Sim-
o can be shown to be r u l e d out by entirely the s a m e a r g u m e n t
L 3 = 4,
Finally,
L
12k+9
4k+3 f L 4 k + 3
The 6 in the b r a c k e t may be e i t h e r 3 or 1.

b e r s L, 7 0 a r e divisible by 3 , it follows that
J
9
4k+2
relatively prime.
t o r s m u s t be such.
o r 2.
But s i n c e only Lucas n u m L . , _ and ~L2A, ,' + 3 a r e

4k+3
4k+3
T h e r e f o r e , if L-j 2 k+9 is-to be a p e r f e c t s q u a r e , both f a c It i s c l e a r that L ^ . o
F o r other v a l u e s ,
L-.^,?
p e r f e c t s q u a r e for k = 1
3.k , 3k + 1 o r 3k' + 2 with kf ^ 1 .
k equals e i t h e r
But this gives us Lucas n u m b e r s
is not
?,
, and L i o u F + l 1
respect-
ively and it has a l r e a d y been shown that t h e s e cannot be s q u a r e s .

Thus the only s q u a r e s in the Lucas sequence a r e L,1 = 1 and L 3 = 4.
c
L e g e n d r e f s symbol.
LATTICE PATHS MB
FIBONACCI HyfVfBEflS
ROBERT E. GREENWOOD
U n i v e r s i t y of T e x a s
L. Moser and W. Zayachkowski [1] considered lattice paths from (0,0)

to (m9n) where the possible moves were of three types: (i) horizontal moves
from (x,y) to ( x + l , y ) ; (ii) vertical moves from (x,y) to ( x , y + l ) , and (iii)
diagonal moves from (x,y) to (x + l , y + 1). A special case of some interest
arises when m = n.
Consider now a much more restricted set of paths. Require: (a) that the
path be symmetric about the line x + y = n, (b) that prior to arriving or touching the above line that one use only horizontal and diagonal moves (and symmetry
after now requires that vertical and diagonal moves be used to arrive at .(n,n)),
and (c) that all of the paths be "below" or on the line y = x (also required by
the previous conditions).
For small values of n one can enumerate the possible paths.
n = 1, one need only consider the three points
Thus for
(0,0), (1,0) and (1,1), and
there are two paths as pictured in Fig. 1.
For n = 2, there will be three
paths. For n = 3 there will be five paths.
See Figs. 2 and 3, respectively.
This suggests that the collection of path numbers may be closely related
to the Fibonacci sequence, with appropriate renumbering to bring the two s e quences into step. Thus s letting h(n) be the number of paths for (0,0), (n,n)
case, one has the tabulation
n
h(n)
1 2 3
2 3 5
Also, beginning at (0,0), there are only two initial moves 9 to .(1,0) and to
(1,1). Due to symmetry imposed by requirement (a) the last move in the path
is also determined so that one has the choices schematically portrayed in Fig. 4.
The two path schemes depicted in Fig. 4 are mutually exclusive and collectively they exhaust all of the allowable paths satisfying conditions (a), (b) and (c)."
Hence, h(n) = h(n - 1) + h(n - 2) for n = 3?49 . Thus the sequence of path
numbers is a Fibonacci sequence with appropriate relabelling and identification
of h(0) and h(-l) as unity.
13
14
LATTICE PATHS AND FIBONACCI NUMBERS
[Feb. , 1964]
It is also possible to "count" the paths so as to have

/n + 1 - k^
h(n) =
^
*0
but the grouping of paths with summation index k seems slightly artificial and
lends little or nothing to the general theory
Figure 1
JB
(two cases)
-#
Figure 2
/*
#-#
JT
(three cases)
/*~~^ #
r #
F i g u r e s (five cases, smaller scale)
(n,n)
(n - l , n - 1 ) . /
/
/
/
/
/
(0,0)
(n, n - 1)
iyVa.:
/
t 4(i,o)
(n-!,()) J
'(n,l)
>(n 0)
'
'Vfti.n)
( n - l , n - 1)/
Z
<,
(n.n-J)
( 0 , 0 ) / ( 1 , 0 ) (n-1,0)*
(ns0)
There are h(n - 1) paths from (1,0)
There are h(n - 2) paths from (1,1)
to (n,n - 1) inside dashed triangle.
to (n - 1, n - 1) insided dashed
triangle.
Figure 4
REFERENCE
1. L. Moser and W. Zayachkowski, "Lattice Paths with Diagonal Steps,"
Scripta Mathematical Vol. XXVI, No. 3, pp. 223-229.
A NOTE ON FIBONACCI NUMBERS

L.
CARLITZ
Duke U n i v e r s i t y , Durham, N. C.
We shall employ the notation

ufl = 0S u< = 1,
u , _. = it
v 0 = 2,
v n+1
= v
n+1
' n +' v* n --l
= 1,
Vl
+ u
.,
(n > 1)
(n => 1) .
Thus
n
(1)
a
n
,>n
- B
or - /3 * n
n ,
nn
p
where
1 + *s/5
^ = r
, /3
The f i r s t few v a l u e s of u , v
u
n
3
2
1 - \/5
a + jS = 1, orjS = - 1
follow.
4
3
7
5
5
13
21
34
11
18
29
47
76
11
12
55
89
144
123
199
10
322 !
n
It follows e a s i l y from the definition of (1) that
(2)
u = u . , n u, + u , u, ,
n
n-k+1 k
n-k k-1
(3)
= u
n
, , n v, + u , v, .
n-k+1 k
n-k k-1
(n > k > 1)
'
(n > k > 1)
'
It i s an i m m e d i a t e consequence of (1) that

(4)
(5)
(6)
v
V
mk
u
2mk
'
k|v(2m-l)k
'
* Supported in part by National Science Foundation Grant G16485.
15
,
'
16
[Feb.
where m and k are arbitrary positive integers. It is perhaps not so familiar

that, conversely 5
(4)'
u k | un = * >
(5)!
u k J u n = = > n = 2mk
(6)?
v, I v
=^>
n = mk
(k > 2)
(k > 1)
n = (2m - l)k (k > 1)
K i n
These results can be proved rapidly by means of (1) and some simple results about algebraic numbers. If we put
(7)
n = mk + r
(0 < r < k)
then
n
a
r . mk
nn
n mk, , n m k , r
- p = a (a
- (3 ) + ft (a
nr.
- ft )
so that
^mk u
un = a r umkT +, B
^
r
If u, j u
R(N/5),
it therefore follows that u, p
u . Since ft is a unit of the field
Ujlu , which requires r = 0. This proves (4) ! .
Similarly if
n = 2mk + r
(0 < r < 2k)
then
u
Hence if v. j u
r
2mk
= a u 0 , + HR
u
2mk
r
it follows that v, I u .
If then r > 0 we must have r > k
and the identity

(a - /3)u r = a0'
gives v, I v
, , . which is Impossible.
vk -
P\_k
The proof of (6)? is similar.
1964J
17
If we prefer, we can prove (4) f , (5) f , (6)f without reference to algebraic

numbers.
For example if u , | u , then (2) implies u, I u
and u,
are relatively prime we have u, I u
,.
, u.
Since u,
Continuing in this way we
get u, I u , where r is defined by (7). The proof is now completed as above,

In the same way we can prove (5)* and (6) f .
In view of the relation
(8)
%
'
u0 = u v
2n
n n
it is natural to ask for the general solution of the equation

(9)
un = u m v k
(m > 2, k > 1) .
It is easily verified, using (1), that (9) can be replaced by
<10>
n = Um+k
(-^Sn-k
<m k>
or
(ID
un = u m + k - (-l)kuk_m
(k > m)
Now the equation

(12)
u r = ug + ut
(s > t > 1)
is satisfied only when r -
1 = s = t + 1. Indeed if 1 < t < s - 1, then

u+u, < u + u 1 = u
s
t
s
s-1
l 1
s+1
.
'
so that (12) is impossible; if t = s - 1, then clearly r = s + 1. If t = 1 in

(12) we have the additional solution r = 4, s = 3.
Returning to (10) and (11) we first dispose of the case m - k = 1.
k even (10) will be satisfied only if m + k = 3, which implies k = 1; for
odd we get n = 2 , m + k = 3
or n = 3 , m + k
= 4,
For
k
which is impossible.
Equation (11) with k - m = 1 is disposed of in the same way.
18
[Feb,

We may t h e r e f o r e a s s u m e in (10) and (11) that jmk| > 1.
Then if k i s
even, i t i s evident f r o m the r e m a r k concerning (12) t h a t (10) i s impossible.
If
,k i s odd t we have,-,
m+k
so that k = 1, m = ne
r..
m-k
As for ( l l ) f if m i s odd we get

u
= u
n
which i s impossible,,
+ U
,, + u,
m+k
k-m
However 9 if m i s even s we get

' u
, , - u + u,
m+k
n
k-m
'
so that m + k = n + 1 = k - m + 2; this r e q u i r e s
m = 1, k = n,
This c o m p l e t e s the proof of

T h e o r e m 1, - The equation.
u
n
=. u v,
m k
(m > 2 9 k > 1)
h a s only the solutions n = 2m = 2k.

The l a s t p a r t of the above proof s u g g e s t s c o n s i d e r a t i o n of the equation
(13)
un = vk
(k > 1) .
Since (13) i s equivalent to
u
n
= u, ,- + u t n
k+1
k~l
it follows at once that the only solution of (13) i s n = 4, k '= 2.

The equation
(14)
un = v m v k
(m =>k > 1) ,
i s equivalent to
(15)
u = v
, + (-1) v
,
v
;
m+k
m-k
If k i s even it is c l e a r that n > m + k; indeed since %^+k
x
we m u s t have n > m + k + 1.
Then (15) i m p l i e s
= U
m+k+l
m+k-l
1964]

U
m+k+2 -
m+k+l
m+k~l
19
m-k
which simplifies to
(16)
v
'
,, o < v
,
m+k-2
m-k
If m = k s (16) holds only when m = 2; however this does not lead to a solution
of (14).
If m > k, (16) may b e w r i t t e n as
< U
m+k-2
, . - + 1 1
m~k+l
which holds only when m = 4 ?
< U
m-k-1
m-k
k = 2.
If k i s odd ? (15) b e c o m e s
(17)
x
u + v - t = v ,,
n
m-k
m+k
If m = k t h i s r e d u c e s to
u
+ 2 = u01'
+ u n l . 9,
2k+l
2k-1
which I m p l i e s 2k - 1 = 3 , k = 2 If m = k + 1 (17) g i v e s '
which Is c l e a r l y impossible*
11
2k'.
F o r m > k + 1 we get
, 1 . 1
m+k+1
s o t h a t n < m + k + 1.
..!=.u2k+2
^1
m+k-1
2 U
m-k
Since
m+k
we m u s t have n = m + k + 1.
2u
m-k
<
m+k+l
Hence (17) b e c o m e s
m-k
m+k-l
m+k~l
20
[Feb.
as we have seen above, this implies

m -i k = 2, m + k - 1 = 4
so that we do not get a solution.

We may state
Theorem 2. The equation
u
n
= v v,
m k
(m ^ k > 1)
'
has the unique solution n = 8, m = 4, k - 2.

It is clear from (4)f that the equation
(18)
u n = cu k
(k > 2) ,
where c is a fixed integer >1 is solvable only when k| n. Moreover the number of solutions is finite. Indeed (18) implies
cu, ^ u 01 > u, v, . c ^ v, ;
k
2k
kk'
k '
moreover if n = rk then for fixed k, r is uniquely determined by (18).
This observation suggests two questions: For what values of c is (18)
solvable and, secondly, can the number of solutions exceed one? In connection
with the first question consider the equation
(19)
u n = 2u k
(k > 2)
Since for n > 3

2u 0 4
1964 J
3u
n - 3 < Un
3u
+ 2u
n-3
n-4 <
3u
n-2
it follows that the equation

(20)
u n = 3u k
(k > 2)
has no solution.
Let us consider the equation
(21)
x
u = u u,
n
m k
(m > k > 2)
;
*
We take
un = u n-m+1
, - u m + un-m u m-1
so that
u
,-u
 mQ Then clearly (21) is impossible,
For the equation
(22)
vn = u m v k
(m > 29 k > 1) ,
we use
v
= u v
,- + u -v
m n-m+1
m-1 n-m
Then
u
mVm+l
<
mvn-m+2
so that (22) is impossible.

This proves
Theorem 3. Each of the equations (21), (22) possesses no solutions.
21
22
[Feb.
Consider next the equation

(23)
v
'
. v
= v v,
m k
(m > k > 1) . "

'
This i s equivalent to
v = v M + (-l)kv
,
l
;
n
m+k
m-k
(24)
v
F o r k even, (24) i s obviously i m p o s s i b l e .

=
m+k
which r e q u i r e s m + k = n + l = m - k + 2
T h e o r e m 4a
F o r k odd we m a y w r i t e
,
m-k
.
'
so that k = 1.
This p r o v e s
The equation (23) p o s s e s s e s no solutions.
The r e m a i n i n g type of equation i s

(25)
x
'
= u u,
m k
> k > 2)
'
5v = v ^ + v( ~ l ) k v
,
n
m+k
'
m-k
x(m
T h i s i s equivalent to
(26)
K ;
Clearly n < m + k.
Then since
m+k
5v
m+k-4
3v
m+k-5
(26) i m p l i e s
(27)
^ ;
5v
= 5v
n
+ 3v _,_, . + ( - l ) k v
,
v
m+k-4
m+k-5
'
m-k
J_1
Consequently n > m + k - 3, while the r i g h t m e m b e r of (27) Is l e s s than
5v
This evidently p r o v e s
m+k-4
4V
m+k-5 <
5v
m+k-3
'
1984]

T h e o r e m 5.
23
The equation (25) p o s s e s s e s no solution.
Next we d i s c u s s the equations

(28)
'
(29)
v
'
+ u2 = u2
n
k
(0 < m -= n)
'
+ v2 = v2
(0 ^ m < n)
;
m
n
k .
We shall r e q u i r e the following

Lemma,
The following inequalities hold.
(30)
- ^
n
> |
(ns
(31)
(D
2)
> 3)
n
Proof.
Since u
< 2u - for n > 2 9 we have

n~l
*
Vi
n
n-r
n
The proof of (31) i s exactly the s a m e .

Returning to (28) it i s evident that
u 2 < u,2 < 2u 2
n
k
n
,
'
so that
u
< u, n and by the l e m m a
u,
Since *f2 < 3 / 2 ,
>
,.,.
>
n+1 "
we have a contradiction.
The l e m m a r e q u i r e s that n > 2 or
3
2 n
The s a m e a r g u m e n t applies to (29).
3 but t h e r e i s of c o u r s e no difficulty about
24
[Feb.
the excluded values. This proves

Theorem 6. Each of the equations (28), (29), possesses no solutions.
More generally, each of the equations
r
r
r
um + un = u,k
v(0
vm + vn = v,k
x(0
< m < n)' >9

^ m < n)'
9,
where r ^ 2 has no solutions.

Remark.
The impossibility of (29) can also be inferred rapidly from the
easily proved fact that no v
is divisible by 5. Indeed since
a5 = /35 = |
(mod's/5)
it follows that
V-5
so that v _ ^ v
= a
n+5 , nn+5 _ 1 , n , Jti.
2{a
+ P
&)
2 Vn
(m
,' /=.
d ^5) '
(mod^5). Moreover none of v 0 , v l s v 3 , v4 is divisible by 5.
The mixed equation

(32)
x
'
v m2 + v 2n = u 2k
(0
v < m < n);
has the obvious solution m = 2, n = 3, k = 5; the equation

(33)
+ v
has the solution m = 4f n = 3 ,

Clearly (32) implies
{m >
0)
k=5.
v n < \x,k < vn
N/2
This inequality is not sufficiently sharp to show that (32) has no solutions although it does suffice for the equation
r
r
+ v
r
=
Ui
1964 ]
with r
25
sufficiently l a r g e .
However (32) i s equivalent to
<34>
(-V2
+ v
2n + ^
= i <v2k " (" 1 ) k 2 >
If m + n EE 1 (mod 2), this r e d u c e s to
2m
2n = I i V 2k " ^
T h e r e i s no l o s s in generality in a s s u m i n g k > 5.
2k =
5v
2k-4
3v
2k-5
Then since
'
we get
v0
2m
o
= V O 1 A + | { 3L v 0 1 _ - ( - l ;) k 2 |j .
2n
2k-4
5
2k-5
Since m < n and
|{
3 v
2k-5-
"1>k2} <
2k-5
we m u s t have 2n = 2k - 4 and
5v
2m =
3v
2k-5 "
~1>k2 =
6v
2k-7
3v
2k-8 " ^
It i s t h e r e f o r e n e c e s s a r y that 2m = 2k - 6 and we get

5v 0
= 6v 0
, + 3v Q
+ (-l)m2
2m
2m-l
2 m - 20
'
which simplifies to
Hence m = 2 s k = 5
n = 3 (a solution of (22)).
Next if m = n (mod 2),
(34) r e d u c e s to
,
*
26

v0
+ v 0 + K( - l; ) n 4 = | l{v 0 1 2m
2n
5
2k
[Feb.
( -; l ) k 2 |
J
and as above we get
<35>
2m+
2n + ^
= V 2k-4
It Is n e c e s s a r y that 2n = 2k - 4,
(36)
Clearly
2m < 2k - 6.
3v
>
= 3v2k_5 - (-l)k2
If 2m < 2k - 6 we get
2k-5"
'2k-6
which Is not p o s s i b l e .
I i 3v 2k-5 " ^
so that (35) r e d u c e s to
(-l)m20
5v2m+
- ^
5V
2k-7+ ^
m 2
2k-
Thus 2m = 2k - 6 and (36) b e c o m e s
5v 0
+ v( - l ;) m 20 = 3v 0
+ K( - l ) m 2
2m
2 m + ln
'
This r e d u c e s to
v0
. = (-1)
2m-4
which i s satisfied by m = 5.
18
'
,
'
Then k = 8, n = 6 but this does not lead to a
solution of (32).
This c o m p l e t e s the proof of
T h e o r e m 7.
The equation
v
+ v^ = u /
n
k
(0
< m < n)
x
'
has the unique solution m = 2, n = 3, k = 5.

The equation
1964]
(37)
+ v2 = u2
m
n
k
can be t r e a t e d in a l e s s tedious m a n n e r .
27
(m > 0)
v
Suppose f i r s t that v
vr
 n + 2
then
u2 >
k
u 2 = v(2u , - + u 7)2 = 2(u

, - + u 1 ;) 2 + 2 u 2 i 1 + 2 u M u
.
x
n+3
n+1
n
n+1
n-l
n+1
n+1 n - 2
+ u 2 - u 2 ., >
n
n-1
2v2
n
'
so that (37) i s c e r t a i n l y not satisfied. Since k > n + 1 it follows that k = n + 2.

Thus (37) b e c o m e s
u 2 = u2 0 - v 2 m
n+2
n.
(38)
v
a s i s easily verified.
2 >
m
c o n t r a d i c t i n g (38).
3 ( u 2 - u2 n )
n
n-1'
If m > n + 2 then
u2
n+2
= (2u
v
+ u
2
)2 > 3 (u
- u2 1 ); J,
v
n - n1 ;
n
n-1
Since for n > 3
3(u 2 - u2 , ;) - u 2 = 2 u 2 - 3u2 , > | u 2 1 - 3u2

> 0
Vn
n-l
n
n
n-1
2 n-1
n-1
it follows that m > n.
Thus m = n + 1 and (38) b e c o m e s

2
u2 , = 3(u
- u 2 , );
x
n+1
n
n-1
This i m p^ l i e s u n + u n - 1- = 3,' n = 3 5' which leads to the solution n = . 3 ,? m = 4.J

k = 5 of (37).
As for the excluded values n = 1, 2 it i s obvious that they do
not furnish a solution.

T h e o r e m 8.
This p r o v e s
The equation
28

u
+ v2 = u2
.n
k
has the unique solution m = 4, n = 3 ,
[Feb.
1964]
(m
> 0);
x
k=5.
The Fibonacci Association invites Educational Institutions to apply

for Academic M e m b e r s h i p in the Association.
s c r i p t i o n fee i s $25 annually.
The m i n i m u m s u b -
(Academic M e m b e r s will r e c e i v e two
copies of each i s s u e and will have t h e i r n a m e s l i s t e d in the J o u r n a l . )
REQUEST
The Fibonacci Bibliographical R e s e a r c h Center d e s i r e s that any r e a d e r
finding a Fibonacci r e f e r e n c e , send a c a r d giving the r e f e r e n c e and a b r i e f d e s c r i p t i o n of the contents.
P l e a s e f o r w a r d all such information to:
Fibonacci Bibliographical R e s e a r c h C e n t e r ,
Mathematics D e p a r t m e n t ,
San J o s e State College,
San J o s e 9 California
NOTICE TO A L L SUBSCRIBERS!!!
P l e a s e notify the Managing Editor AT ONCE of any a d d r e s s change.
The P o s t
Office D e p a r t m e n t , r a t h e r than forwarding m a g a z i n e s m a i l e d t h i r d c l a s s , s e n d s

t h e m d i r e c t l y to the d e a d - l e t t e r office.
U n l e s s the a d d r e s s e e specifically
re-
q u e s t s the Fibonacci Q u a r t e r l y b e f o r w a r d e d at f i r s t c l a s s r a t e s to the new a d d r e s s , he will not r e c e i v e it.
(This will usually c o s t about 30 cents for
first-
c l a s s postage.) If p o s s i b l e , p l e a s e notify us AT LEAST THREE WEEKS PRIOR

to publication d a t e s :
F e b r u a r y 15, April 15, October 15, and D e c e m b e r 15.
RENEW YOUR SUBSCRIPTION!!!
SOME HEW FIBONACCI IBEPfTITiES

VERNER
E. H O G G A T T , J R . and M A R J O R I E
San J o s e S t a t e C o l l e g e , San J o s e ,
BICKNELL
California
In this p a p e r , s o m e new Fibonacci and L u c a s identities a r e g e n e r a t e d by

matrix methods.
The m a t r i x
R = I 0
Vi i
s a t i s f i e s the m a t r i x equation
R 3 - 2R 2 - 2R + I = 0
Multiplying by R
yields
R n + 3 - 2Rn+2 - 2Rn+1 + Rn = 0
(1)
It has been shown by B r e n n a n [1 ] and a p p e a r s in an e a r l i e r a r t i c l e [ 2]

and as E l e m e n t a r y P r o b l e m B-16 in this q u a r t e r l y that
Rn
(2)
w
where
R
F2
n-1
= ( 2F F ,
n n-1
F2
n
-F
n-1 n
F*
- F -F
n+1
n-1 n
F F ^
n n+1
F2
n
2F F 1
n n+1
F2
n+1
is the n Fibonacci n u m b e r ,
n
n+3
By the definition of m a t r i x addition^ c o r r e s p o n d i n g e l e m e n t s of R
,
, R
and R
m u s t satisfy the r e c u r s i o n formula given in Equation (1).
That i s s for e x a m p l e ,
29
30
SOME NEW FIBONACCI IDENTITIES
[Feb.
F2
- 2 F 2 J Q - 2F2
+ F2 = 0
n+3
n+2
n+1
n
and
F
F
-2F
F
- 2F
F
+ F F
n+3 n+4
n+2 n+3
n+1 * n+2
n n+1
=0
u
'
Returning again to
R 3 - 2R 9 ~ 2R + I = 0
this equation can b e r e w r i t t e n a s

(R + I) 3 = R3 + 3R2 + 3R + I = 5R (R + I )
In g e n e r a l , by i n d u c t i o n it can be shown that

RP(R
(3)
I)2n+1 = 5 n R n + P ( R + I )
Equating the e l e m e n t s in the f i r s t r o w and t h i r d column of the above m a t r i c e s ,

by m e a n s of Equation (2), we obtain
2n+l
V"*
(4)
I
i=0
/ 0 , ., \
/ 2n + 1 \
x
i + p= 5 n F 2(n + P ) + l
It i s not difficult to show that the Lucas n u m b e r s and m e m b e r s of the

Fibonacci sequence have the r e l a t i o n s h i p
L 2 - 5F 2 = v( - l ); n 4
n
n
Since a l s o
2 n
/2n
l\
.+
we can d e r i v e the following s u m of s q u a r e s of L u c a s n u m b e r s ,
1964]

2n+l
. \
/ o
2
I+p
-&+1 "
F
2(n+p)+l
i=0
by substitution of the p r e c e d i n g two identities in Equation (4).
Upon multiplying Equation (3) on the r i g h t by (R + I), we obtain
R P (R + l ) 2 n + 2 = 5 n R n + P ( R +
(5)
Then, using the e x p r e s s i o n for R
I)2
given in Equation (2) and the identity
k =
k-i
k+i
we find that
(R
n+1
R )(R + I)
2n-1
2n
( 2F2n
2F2n+1
2n+1
. F 2n+2
2n
2L
L
2n+3
2n+1
2n+2
2n+2
2L
2n+3
2n+4
L
2L
2n+l
2n+2
2F2n+2
2n+3
Finally 8 by equating the e l e m e n t s in the f i r s t row and t h i r d column of the

t r i c e s of Equation (5) 9 we derive the two identities
2n+2
i=0
'2n+2
v
and
2n+2
i=0
i+p -
2(n+p)
32
[Feb. 1964]
By s i m i l a r s t e p s , by equaling the e l e m e n t s a p p e a r i n g in the f i r s t r o w and

second column of the m a t r i c e s of Equations (3) and (5), we can w r i t e the a d d i tional identities 9
2n+1
'2n
iC
i=0
l\
F
F
i-l+p
5nF
i+p
2(n+p)
and
211+2
/' 22 nn ++ 2 \.
=
SnT
i+p
2(n+p)+l
REFERENCES
1.
F r o m the unpublished notes of T e r r y B r e n n a n .
2.
M a r j o r i e Bicknell and V e r n e r E Hoggatt 9 J r . ,
"Fibonacci M a t r i c e s and
Lambda F u n c t i o n s / ' The Fibonacci Q u a r t e r l y , 1 (1963) ? A p r i l s pp 47 520
TWO CORRECTIONS, VOL, 1, NO. 4

P a g e 73: In p r o p o s a l B - 2 6 , the l a s t equation should r e a d
B v(x) = v(x + 1) B -(x)f + b -(x)
n '
' n-V
n-lv ;
P a g e 74: In p r o p o s a l B - 2 7 , the line for cos n< should r e a d
N
c o s n 0 = P n (x) = y
A.nxn+2"2j(N
[(n+2)/2]
pi
RENEW YOUR SUBSCRIPTION!!!
PRIMES WHICH ARE FACTORS OF ALL

FIBONACCI SEQUENCES
BROTHER
U.
ALFRED
St. M a r y ' s C o l l e g e ,
California
In studying the Fibonacci and Lucas sequences, one of the striking differences observed is the fact that ALL primes are factors of some positive
term of the Fibonacci sequence while for the Lucas sequence many primes are
excluded as factors e This difference raises some Interesting questions regarding Fibonacci sequences in general.
(1) For a given Fibonacci sequence, how do we find which primes are factors
and which are non-factors of Its terms?
(2) Are there certain primes which are factors of all Fibonacci sequences?
It Is this latter question which will be given attention in this paper,
We are considering Fibonacci sequences in which there is a series of
positive terms with successive terms relatively prime to each other.
For any
sequence we can find two consecutive terms a G9 b > 0, a < b, and take
these as
f0 = a9
fj = b
the defining relation for the sequence being

fn+1
^ = fn + fn-1, s,
(n ^ 2);
The particular sequence with a = 0 and b = 1 is known as the Fibonacci s e quence and will have its terms designated by F 0 = 09 Ft = 1, and so on.
Theorem: The only Fibonacci sequence having all primes as factors of
some of Its positive terms is the sequence with a = 0 and b = 1.
Proof: Since zero Is an element of the sequence, the fact that all primes
divide some positive terms of the sequence follows from the periodicity of the
series relative to any given modulus.
To prove the converse, we note that each sequence is characterized by a
th
is the n term of the sequence,
quantity D = b 2 - a (a + b). For if f

f
- F
,b + F 0 a
n-1
n-2
33
34 PRIMES WHICH ARE FACTORS OF ALL FIBONACCI SEQUENCES
[Feb,
Then
f2 - f
f L 1 = (F
, b + F 0 a ) 2 - (F
b + F c a );(vF b + F ,a)
l
v
n
n-1 n+1
n-1
n-2 '
n-20
n-3
n
n-1 '
which equals
2
2
b 2 (F
- F 0 F ); + ab(F , F 0 - F F Q ); + a2 (F
- F QF ,)
v
v
n-1
n-2 n
n-1 n-2
n n-3
n-2
n-3 n - 1 '
or
(~l) n (b2 - ab - a 2 )
so that the values are successively +D and -D.
Now D is equal to 1 in the case of the sequence 0 , 1 , 1 , 2 , 3 , and in
no other Fibonacci sequences.
a2 increases with b0
For if a is kept fixed, the quantity b(b - a) -
Therefore its minimum value is found for b = a + 10
But then b(b - a) - a2 becomes a + 1 - a20
Now if a = 0,1 or 2, [a + 1 - a2|
= 1 and we have the Fibonacci sequence. If a ^ 3, | a + l - a 2 | ^ 5 .

Thus, apart from the Fibonacci sequence properly so-called,
Furthermore, D must be odd if a and b are relatively prime.
f
D > 1.
Hence if
= 0 modulo some prime factor p of D, we would then have

f
n-l fn+l
< m o d P>
n " fn-l fn+l = ^ ^
from the relation
so that either f
or f
- = 0 (mod p) Thus two successive terms of the
series would be divisible by p and consequently all terms would be divisible

by p which would lead to the conclusion that p | (a,b), contrary to hypothesis,
Therefore, the only Fibonacci sequence having all primes as divisors
one or the other of its terms is the one Fibonacci sequence with a zero element,
namely: 0 , 1 , 1 , 2 , 3 , 5f 8,13, .
CONGRUENTIAL FIBONACCI SEQUENCES
For a given prime modulus, such as eleven, there are eleven possible
residues modulo 11: 0,1,2, 3, , 10. These may be arranged in ordered pairs
1964] PRIMES WHICH ARE FACTORS OF ALL FIBONACCI SEQUENCES
35
repetitions being allowed in l l 2 or 121 ways. Each such pair of residues can
be made the starting point of a congruential Fibonacci sequence modulo 11,
though of course various pairs will give rise to the same sequence.
The one
pair that needs to be excluded as trivial is 0 - 0 since all the terms of the s e quence would then be 0 and we have assumed throughout that no two successive
terms have a common factor. Hence there are 120 possible sequence pairs. A
complete listing of these congruential sequences modulo 11 is displayed below.
(A)
(B)
(C)
(D)
10
(E)
(F)
10
(G)
(H)
(I)
9
10
(J)
(K)
(L)
(M)
That all
2
4
10
9
1
2
10
3
4
10
10
10
5
8
10
10
2
6
8
10
7
possible sequence-pairs are covered is shown in the following table
where the number in the column at the left is the first term of the pair and the
number in the row at the top is the second.
4
5
2
3
1
0
E
D
B
C
A
X
0
8
H
9
I
10
J
H.
6
7
E
I
F
M
C
H
B
E
C
F
H
9
10
J
A
B
G
L
K
C
E
H
B
I
M
J
D
I
J
D
J
36
PRIMES WHICH ARE FACTORS OF ALL FIBONACCI SEQUENCES
[Feb.
We shall now c o n s i d e r v a r i o u s c a t e g o r i e s so a s to c o v e r all p r i m e s .

(A) p = 2
If e i t h e r a or b i s even^ 2 i s a factor of t e r m s of the s e r i e s ; If both
a r e odd 9 then a + b = 0 (mod 2), T h u s , 2 Is a factor of all Fibonacci sequences,"
(B) p_
Since 5 Is not a factor of t e r m s of the L u c a s s e r i e s , It cannot be a d i v i s o r of all Fibonacci s e q u e n c e s .
(C) p = lOx + 1
F o r p of the form 1 0 x l , the p e r i o d h(p) for any Fibonacci sequence
i s a d i v i s o r of p - 1.
Since t h e r e a r e p 2 - 1 sequence p a i r s of r e s i d u e s , the
n u m b e r of congruential s e q u e n c e s modulo p would have to be
p2 - 1
or
p + 1
But s i n c e t h e r e a r e only p - 1 r e s i d u e s o t h e r than z e r o , sequence t r i p l e s

a-0~a can only be p - 1 in n u m b e r ,
Thus t h e r e cannot be one p e r s e q u e n c e .
Hence no p r i m e of the f o r m 10 x . 1 c a n b e a d i v i s o r of all Fibonacci s e q u e n c e s .

(D) p = 10 x 3
F o r p of the f o r m
10 x + 3, the situation Is a s follows:
(1) The p e r i o d is a factor of 2p + 2.

(2) 2p + 2 i s divisible by 4*
(3) The p e r i o d contains all power of 2 found in 2p + 2.
(4) The p e r i o d i s the s a m e a s the p e r i o d of the Fibonacci s e q u e n c e , F
Accordingly 9 if the p e r i o d Is l e s s than 2p + 2 9
.[1 ]
it will a l s o be l e s s than
p - 1 and hence a s before t h e r e will not be enough sequence p a i r s with z e r o s

to c o v e r all the s e q u e n c e s .
Thus a n e c e s s a r y condition Is that the p e r i o d be
2p + 2 if a p r i m e Is to b e found a s a factor of all Fibonacci s e q u e n c e s ,

Two c a s e s may be distinguished:
2
(a) The c a s e in which the p e r i o d h(p)
= 2 (2r + 1); (b) The c a s e i n which the p e r i o d h(p) = 2A
(2r + 1), m ^ 3.
1964] PRIMES WHICH ARE FACTORS OF ALL FIBONACCI SEQUENCES
37
(a) h(p) = 2 2 (2r + 1)

In this instance, if a sequence has a z e r o at k,
it will also have z e r o s
a t k / 4 , k / 2 g and 3 k / 4 or four z e r o s p e r sequence. The n u m b e r of s e q u e n c e s

is
P2" - 1
2p + 2
P - 1
2
To provide 4 z e r o s p e r sequence t h e r e would have to be
Ê"-""^ =
(P "
zeros,
w h e r e a s t h e r e a r e only p - 1.
m
(b) k = 2
(2r +"1), m ^ 3.
F o r a p e r i o d of this form, if t h e r e is a z e r o at k,
z e r o at
k/29
but not at k / 4 or
3k/4.
t h e r e will a l s o be a
The n u m b e r of z e r o s r e q u i r e d for
(p - l ) / 2 s e q u e n c e s would be
2(p'- l)/2.= p -;1
which i s the exact n u m b e r available. Thus the p r i m e s which divide all Fibonacci
m
s e q u e n c e s a r e p r i m e s of the f o r m 10 x 3 for which 2p + 2 i s equal to 2 (2r
+ 1), m > 3. In other w o r d s ,
p = +3 (mod 10)
p = 2m~1
(2r + 1) - 1 o r p s= - l
(mod 4)
T h e s e c o n g r u e n c e s lead to the solution p = 3,7 (mod 20).

CONCLUSION
The p r i m e s which a r e f a c t o r s of all Fibonacci s e q u e n c e s a r e :
(1) The p r i m e 2
(2) P r i m e s of the f o r m
20k + 3 , 7 ,
having a p e r i o d 2p + 2
38 PRIMES WHICH ARE FACTORS O F ALL FIBONACCI SEQUENCES

LIST O F PRIMES WHICH DIVIDE
A L L FIBONACCI SEQUENCES (p < 3000)
383
787
1327
1783
2383
443
823
1367
1787
2423
463
827
1423
1847
2467
23
467
863
1447
1867
2503
43
487
883
1487
1907
2543
67
503
887
1543
1987
2647
83
523
907
1567
2003
2683
103
547
983
1583
2063
2707
127
587
1063
1607
2083
2767
163
607
1123
1627
2087
2803
167
643
1163
1663
2143
2843
223
647
1187
1667
2203
2887
227
683
1283
1723
2243
2903
283
727
1303
1747
2287
2927
2347
2963
367
REFERENCE
1.
D. D. Wall, "Fibonacci S e r i e s Modulo m 9
>?
The A m e r i c a n
Monthly, J u n e - J u l y , 1960, p . 529.
SOME CORRECTIONS TO VOLUME 1 9 NO. 4

Pages 4 5 - 4 6 :
D = 31 should r e a d ( 2 , 7 ) , (3>8).
T h e r e w a s an o m i s s i o n in the Table of n D f s " a s follows:

D
305
311
319
331
341
349
355
359
D
(1,18) (16,33)
(5,21) (11,27)
(2,19) (7,23) (9,25) (15,32)
(3,20) (14,31)
(1,19) (4,21) (13, 30) (17, 35)
(5,22) (12,29)
(6,23) (11,28)
(7,24) (10,27)
361
379
389
395
(8,25) (9,26)
(1,20) (18,37)
(5,23) (13,31)
(2,21) (17,36)
Mathematical
A FIBONACCI TEST FOR CONVERGENCE

J . H.
JORDAN
Washington State U n i v e r s i t y , P u l l m a n , Wash.
Let g(n) be a non-increasing positive function defined on the positive inOO
tegers.
There are many available tests to determine whether or not
2 g(n)
n=l
converges. It is the purpose of this paper to exhibit a test for convergence
which utilizes the Fibonacci numbers.
THE FIBONACCI TEST
oo
oo
2 g(n) converges if and only if 2 f g(f ) converges, where f

n=l
n=l
the nth Fibonacci number.
OO
Proof: Assume 2 g(n) converges.
n=l
| g(l)
|fig(f2)
| { g ( l ) + g(2)}
|f2g(f3)
\ {g(2) + g(3) + g(4) + g(5)}
- \
f 3 g(f 4 )
\ (g(3) + g(4) + g(5) + g(6) + g(7) + g(8)} ^ \
f 4 g(f 5 )
The sum of all terms on the left side of this array is
2 g(n). The sum of all
terms on the right side of the array is

l^fn-l^fa>
n=2
f
l Z
n^fn>
n=2
Since the left side dominates the right side

o
Assume that
2 g(n) diverges.
n=l
39
2 f g(f ) converges.
n=l
is
40

f2g (f i)
= K(l)
hs <f2)
g(l) + g(2)
fdg (f3)
g (2)
>
g(fQ)
n+A>
[Feb.
+ g(3) + g(4)
+ ... +
g (f n
n+2 "
The sum of all terms on the right side of this array is 2 2 g(n). The sum of
00
all. terms on the left side of this array is

2 g(n) diverges so does 2 f g(f )
n=l
n=l
n == X 00
2 f +1 g(f ) 2 2 ^ gtf )
n=l '
n=l
Since
FURTHER REMARKS
It should be noticed that this result can be generalized to the following:
k
Theorem: If 1 : - c, = [H a ]* where a > 1 and H is a fixed posi00
00
tive constant then
2 g(n) converges If and only if 2 c, g(c, ) converges.

n=l
k=l
The proof is quite similar,
It seems unlikely that this Fibonacci test for convergence will ever be-
come widely used.
To designate some of Its useful qualities the following r e -
sults are exhibited*

a
la
Corollary 1: 2 n converges or diverges as 2 n In n does s a < 0S
n=l
n=2
Proof: Let r be the golden ratio, i.e. 9 r = (1 + 4b)/2 and notice that
f
= [r
/N/S]*.
NOW
00
00
2 n^ converges or diverges as
2 n In r
n=l
n=l
a
does.
00
But n In r = (In r )a which Is approximately (In \l5 f _- ) , So 2 n In r

00
converges or diverges as
2 (in\/5 f ) ' does. Now

n=l
> (ln\/5 f ) a = Y f f""1 (In 45 f f

n=l
n=l
and appealing to the Fibonacci test this converges or diverges as
*~f~x ] isTgreatest integer in x.
n=l
1964]
41
oo
i T 1 (In N/5n) a
^
n=l
does,, which i s e s s e n t i a l l y the d e s i r e d r e s u l t .

This c o r o l l a r y tells us for example that s i n c e the h a r m o n i c s e r i e s gliOO
OO
-1
-1-e
2 n
In n d i v e r g e s and s i n c e
2 n
n=2
n=l
2 n (In n)
converges,
n=2
co
C o r o l l a r y 2.: 2 n
In n c o n v e r g e s or d i v e r g e s as
n=2
v e r g e s then
(n In n)
converges
then
(In In n
n=3
does.
The proof i s quite s i m i l a r .
Corollary j :
j = 3, 4, 5,
a r e likewise p r o v a b l e .
The Fibonacci t e s t i s an effective substitute for the i n t e g r a l t e s t in each

of t h e s e c o r o l l a r i e s .
test.
Consider the following e x a m p l e that i s handled e a s i l y by the Fibonacci

a
Let g(n)
= f
for f
-, < n < f 9. w h e r e a < 0. Thus
& w
m
m-1
m
g(n) = l a + 2 a + 3 a + 5 a + 5 a + 8 a + 8 a + 8 a + 1 3 a +
n=l
Applying the Fibonacci t e s t one obtains
n+1
l (a+1)
l w -Z r - M
r
n=l
n=l
i.i
T
which c o n v e r g e s o r d i v e r g e s as
n=l
/ n+1,a+1
. a+1 .n+1 , , , , . .
2 (r
)
= 2 (r
)
$ out this is a
n=l
n=l
a+1
geometric progression and converges provided r
< 1 or when a < -1.
EXPlOStSHa FIBOSÂGCI HESIDUES

BROTHER U. A L F R E D
St. Mary's College, California
Mathematicians have developed a simple and powerful method of delating

numbers to each other from the standpoint of division,, They say that two numbers a and b are congruent to each other modulo m when the difference a
- b is divisible by m. It can be readily shown that this is equivalent to the
statement that on dividing a and b by m, they will both give the same " r e mainder" which the mathematician calls the least positive residue,
What remainder is obtained when we divided one Fibonacci number by
another? The remainder could of course be zero, but as is well known, zero
is one of the Fibonacci numbers. Do we always obtain a Fibonacci number for
the least positive residue? If not, will we obtain a Fibonacci number if we a l low the use of either the least positive or the least negative residue?
This is the general line of investigation.
In cases in which Fibonacci
numbers are the result, the investigator should seek to find some type of r e g ularity and thus formulate a mathematical theorem. Once this has been done
a proof is a desideratum,,
Just to start the process let us find a few least positive and least negative
residues.
Using F20(6765) as the dividend and various Fibonacci numbers as
divisors, we find that

6765 = 33 (mod 34)
6765 s 0 (mod 55)
6765 = 141 (mod 144)
6765 = 8 (mod 233)

6765 = 356 (mod 377)
6765 = 55 (mod 610)
6765 = 843 (mod 987)

6765 = 377 (mod 1597)
It seems that in some cases the least positive residue is a Fibonacci number
whereas in others apparently it is not. In the latter, we go to the least negative
residue, we apparently get Fibonacci numbers in these cases as well.
Thus
6765 = -21 (mod 377); 6765 = -144 (mod 987).

Those making discoveries in regard to this problem are encouraged to
send their findings to Brother U. Alfred, St. Mary ! s College, Calif., by July
31st so that they may be published in the October, 1964, issue of the Quarterly.
42
TRANSCEHDESITAL HUMBEKS
BASED OH THE FIBONACCI SEQUENCE
DONALD KNUTH
California Institute of Technology, Pasadena, California
A well-known theorem due to Liouville states that if is an irrational

algebraic number of degree n, then the equation
S-
(1)
n+e
has only finitely many solutions for integers p,q, given any e > 0.
There-
fore, an irrational number , for which

(2)
has solutions for arbitrarily large t, must be transcendental.
Numbers of
this type have been called Liouville numbers.

In 1955, Roth published his celebrated improvement of Liouville f s theorem,
replacing "nM by "2" in equation (1). Let us call an irrational number ,
for which
(3)
2+e
his infinitely many solutions for some > 0, a Roth number.
Roth numbers
are also transcendental, and they include many more numbers than the Liouville
numbers.
Let b be an integer greater than 1. Then we define ^,
to be the con-
tinued fraction
(4)
,b
""'' + ?r- +
Theorem: , is a Roth number, hence i|,

43
is transcendental.
44
TRANSCENDENTAL NUMBERS BASED ON

Proof:
F r o m the e l e m e n t a r y t h e o r y of continued fractions, it is well
known that if p / q
is the n
*n m
convergent
to ,,$
fe
h
P
(5)
q 0 = 1,
then
-i /
q
Fm i
In this c a s e ,
[Feb. 1964]
q t = b f
 we have q , - , / q
-* b
/ b
= b
. ~ [(b l ) q y w h e r e <p = .618 i s the golden r a t i o . T h e r e f o r e for l a r g e n we
have approximately
2+cf)
n
and this c o m p l e t e s the proof of the t h e o r e m ,
R e m a r k s , It can be easily shown that the s e t of Roth n u m b e r s i s of m e a s u r e
F o r e x a m p l e , the n u m b e r 2 . b n w h e r e {c }
F
L
*
n=l
nJ
i s a s t r i c t l y i n c r e a s i n g s e q u e n c e of positive i n t e g e r s , i s a Roth n u m b e r ifl i m sup v(c - / c ;) > 2, and it is a Liouville n u m b e r if this l i m sup i s i n n->oo^
n+1/ n
^
finite. In t e r m s of continued f r a c t i o n s , the n u m b e r
z e r o , but it is uncountable.
J,
a1 + , 1
1
a
2
a* +
i s a Roth n u m b e r if and only if

l i m sup (log a i V / log q n )
where q
> 0
a r e the d e n o m i n a t o r s as in the proof of the above t h e o r e m .
The r a p i d c o n v e r g e n c e of (4) allows us to evaluate ,

p r e c i s i o n , e. g ,
R e f e r e n c e to this a r t i c l e on p 9
2 = .70980 34448 61291

^ 3 = ,76859 75625 93155
520
easily with high
STRENGTHENED INEQUALITIES
FOR FIBONACCI AND LUC1S NUMBERS
DOV JARDEN
Jerusalem,
Israel
In a p a p e r entitled "On the G r e a t e s t P r i m i t i v e D i v i s o r s of Fibonacci and

Lucas N u m b e r s " (henceforth r e f e r r e d to a s P ) , published In The Fibonacci
Q u a r t e r l y , Volume 1, Number 3 g pages 15 20 9 I have p r o v e d for the F i b onacci n u m b e r s
F . and the Lucas n u m b e r s

>
F2
F5x+]_
>
5F^
>
L2
(4)
(5)
(41
nx+i
(n
>
L
2,
the following i n e q u a l i t i e s :
x ^
1)
(x * 1)
(n > 2S x > 1)
nx
The aim of this note i s to strengthen (4), (5), and (4*) as follows:
(A)
(B)
> nFnx
x+1
>
n
L11"1
A X
(n > 2, x >
1)
(n > 2 S x > 1)
F o r the proof of (A), (B) we shall u s e the well-known f o r m u l a e

,_.
(C
^5
&
1x
"ii-i
Q
" ^
'
>
1 + N/5
3
n+2
n+1 , n
a = .77- > 77 5 a'
= or
+ a
(D)
= a
+ (-1)
a s well a s the following inequalities:
(E)
^"f-2"1 > n (n 3>

ffn
(F)
(G)
>
Fn
(n >
2)
> Ln
(n 2
2)
45
46
STRENGTHENED INEQUALITIES FOR FIBONACCI AND

Proof of (E) (by induction).
(E ? )
[Feb,
(E) is equivalent to
6 2 n > 7nfsf5
(E f ) i s valid for n = 3.
If (E f ) i s valid for n,
then:
6 2 n + 1 = 6 2 n + 6 2 n > 7iW5 + 7nNT5 > 7nNfs + 7^5 = 7^5(n + 1) .

Proof of (F),
(G) (by induction on n and n + 1),
(F) i s valid for n = 2 , 3 ,
^3
a2
IJL^S
since
3_^5
+ ^5
>
2 + V4
2F8
If
"
a
>
2 F
>
2F
n+l
then a l s o :
n+2
a
n ,
= a
n+1
, _
0 . _,
v
> 2(Fn+ F ^ ) =
_,
2Fn+2
(G) may be p r o v e n analogously, noting that, by a r g u m e n t s employed in the

proof of (F), (G) i s valid for n = 2 , 3 ,
since
6 3 + N/~4
5 ' ~2
=
f a* > - 4 > 4 = L3 .
Proof of (A).
(1) F o r n = 2 we h a v e , by (C):
1964]
2 x+1
LUCAS NUMBERS
=
V5 t a
lK x + 1 -(2
2X+1
-2x+1i
"
^5
> = T
r 2x+1
{a
*X+1)} = | { ^ x + 1 - 2
"
+
47
-2x+1i
r 2X+1
>> 5 ^
_ a
-2X+1T
c-2X+1} = 2 [ ^ ( ^ _ c v - 2 x ^ 2 =
> >
2F2x
(2) For n ^ 3 we have, by arguments employed in the proof of (F),

a n X + 1 => a** = ( a 3 / > 43 > 7
i. e e ,
-
> i .
Hence, by (C), (E):
V i - ^ x + 1 - <-i>n-nX+1> > A f ^ 1 - ^ } =
- i - . f ^ x + 1 = - i . f . 2 n ^ ) n >nFnx .
"5T
"V1
"5C?
Proof of (B). For n ^ 2 we have (n - l)/(n - 1) = n

+n
+
+ 1 ^ n*" 1 ^ (n - l ) * " 1 , whence: n x - 1 ^ (n - l ) x . Hence, by (D), (G), and
x+1
1
noting that (by arguments employed in the proof of (A), part (2)) -a~n
> -
we have:
-nx+1 . nX+1 1 ^
T
nx+l , /1xn _nx+l . nx+l
nx+i
/
nx+l
1 nx+l
a(n-l)x(anx)nrl
\5
Remark.
nx-l / nx\n~l -,
2 . nx.n . 1 , nx.n
^ n - l ^ n - l . > Qpn-l^n-1
nx
In proving the inequalities (A), (B), I was assisted by my son,
Moshe9 who also noted that (B) cannot be strengthened, analogously to (A), to:
L
, > L n.
nx
n x+l
Indeed, for n = 4, x = 1, we have: L 2 = 2207 < 2401 = 7 4

4
= 4 It may also easily be seen, by (C), (D), that

(H)
lim F nx+1
x-^-oo
n
nF
nx
48 STRENGTHENED INEQUALITIES FOR FIBONACCI AND LUCAS NUMBERS

L
lim
nx+1
which shows that, for any given n ^ 2,

x > X,
> nF
> L ; .
nx
By m e a n s of (A), (B), and employing the s a m e r e a s o n i n g a s in the proof
nx+i
t h e r e e x i s t s an X such that, for any
11 1
nx
nx+i
of (3),5 (3*) in P , we have, for the g r e a t e s t p r i m i t i v e d i v i s o r s

L
of L
F?
of F
, the following g e n e r a l i z e d Inequalities:
(J)
> pF9'1
F*
pX-f-i
( p a p r i m e + 5, p ^ 2, x > 1)
pX
(K)
F^ + 1 > Fj x
(L)
L?x+1
P
(x^l)
> LP~2
p"
(p a p r i m e ,
p > 2, x ^ 1) .
SOME CORRECTIONS TO VOLUME 1, NO. 3

P a g e 18: In Equation (4*), r e p l a c e n > 2
by
n > 2,
The l a s t line should r e a d :

. . . for any positive i n t e g e r n > 2,
n > 2,
respectively.
P a g e 17: On line 6, add > to r e a d :
1 + /\/5
2
"
>
1 + f4:
2
"
3
2
Line 8, Equation (7), should be c o r r e c t e d to r e a d :
3
2
On Line 1 1 , add = to r e a d :
P
^
'
1 -^5
2
1
"
- N / 4 =
1
""2
and
ADVANCED PROBLEMS AND SOLOTI0NS

E d i t e d by V E R N E R
E. H O G G A T T ,
JR.
California
Send all communications concerning Advanced Problems and Solutions to

Verner E. Hoggatt, Jr. , Mathematics Department, San Jose State College, San
Jose, California,
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, solu-
tions should be submitted on separate signed sheets within two months after
publication of the problems.
H~29
Proposed by Brother U. Alfred, St. Mary's College,
California.
Find the value of a satisfying the relation

n11 + (n + a) n = (n + 2a) n
in the limit as n approaches infinity.
H - 3 0 Proposed by J. A. H. Hunter, Toronto, Ontario, Canada
Find all non-zero integral solutions to the two Diophantine equations,

(a)
X2 + XY + X - Y2 = 0
(b)
X2 - XY - X - Y2 = 0
H-31
Proposed by Verner E. Hoggatt, Jr., San Jose State College, San Jose,
California
Prove the following:

Theorem: Let a , b , c 9 d be integers satisfying a > 0$ d > 0 and ad be = 1, and let the roots of \2 - X - 1 = 0 be the fixed points of
w = ^44
cz + d
49
50
ADVANCED PROBLEMS AND SOLUTIONS
[Feb,
Then it is necessary and sufficient for all integral n * 0, that a = F 9

,
th
b = c = F 2 , and d = F 2 - , where F is the n Fibonacci number. (Ft
= 1, Fo = 1 and F l 0 = F f 1 + F for all integral n.)
* c
n+z
n+l
n
H-32
Proposed by R. L. Graham, Bell Telephone
Laboratories,
Murray Hill, N. J,
Prove the following:

Given a positive integer n, if there exist m line segments L. having
lengths a. , 1 < a. ^ n9 for all 1 ^ i ^ m s such that no three L. can be
l
th
used to form a non-degenerate triangle then F ^ n5 where F
is the m
Fibonacci number.
H-33
Proposed by Malcolm Tallman, Brooklyn, N. Y.
If a Lucas number is a prime number and its subscript is composite, then

the subscript must be of the form 2 , m ^ 2ffl
SOLUTIONS
A TOUGH PROBLEM
H-l
Proposed by H. W. Gould, West Virginia University,
Find a formula for the n
Morgantown, W. Va.
non-Fibonacci number, that i s , for the sequence
4, 6,7, 9,10,11,12,14,15,16,17,18,19,20,22,23,.
(See paper by L, Moser
and J. Lambek, American Mathematical Monthly 9 voL 61 (1954), pp. 454-458.)

A paper by the proposer will soon appear in the Fibonacci Quarterly,
which will discuss this problem.
A WORLD-FAMOUS PROBLEM
H-2
Proposed by L. Moser and L. Carlitz, University of Alberta, Edmonton,

arid Duke University, Durham, N.C.
Alberta,
Resolve the conjecture: There are no Fibonacci numbers which are integral squares except 0 , 1 , and 144.
See n Lucas Squares/ 1 by Brother U. Alfred in this issue. A discussion
by J. H. E. Cohn on Fibonacci Squares will be in the next issue.
1964]
51
AN UNSOLVED PROBLEM
H-15
Proposed by Malcolm H. Tallman, Brooklyn, N. Y.
Do t h e r e e x i s t i n t e g e r s N l 5 N 2?
and N3 for which the following e x p r e s -
sions cannot equal other Fibonacci n u m b e r s ?

(i)
x
'

(iii)
F3 - F2 F
- F3
n
n m
m
F
n + F n F m + F n F m
F2 - 3 F 3
m , n > N1 ,
l
'
*
m
n
N
> * 2 .
m ? n > No .
No d i s c u s s i o n of any kind h a s b e e n r e c e i v e d on this p r o b l e m .

AN INSPIRING P R O B L E M
H - 1 7 Proposed by Brother U. Alfred, St. Mary's College,
11
Sum
California
It5Fk
k=l
(Editorial Comment: T h e r e will be t h r e e different a p p r o a c h e s to the solution

of the g e n e r a l c a s e of the above p r o b l e m which will a p p e a r soon in the Fibonacci
Quarterly.)
'Solution by Joseph Erbacher and John Allen Fuchs, University of Santa Clara, Calif.
Let L(E) = (E 2 - E - I ) 4 = . 5 0 a - E
x
that E F k = F k ..
w h e r e E i s the l i n e a r o p e r a t o r s u c h
Then L ( E ) k F k = 0. (This follows from a r e s u l t of J a m e s
A. J e s k e P " L i n e a r R e c u r r e n c e Relations - P a r t I, f f Fibonacci Q u a r t e r l y , April*

n
8
1963, p. 72, Equation (4.8).) Let S = ^
k3Fk
Since . ^ a. = 1 , S =
1 a.S = I (a. nzl
k 3 F k + a . lz [j 3 F - (n + j) 3 F , ]) = R + T 9 w h e r e R
l
1
K
J
n j
i=0
i=0^ k=i+l
j=l
J
i s the f i r s t double s u m m a t i o n and T i s the second double s u m m a t i o n . R e v e r s n 8
ing the o r d e r of s u m m a t i o n in R s we have R = 2 2 a.(i + k) 3 F . , , Since
8
8
k=li=0
2 a.(i + k) 3 F
= 2 a . E x k 3 F k = L ( E ) k 3 F k = 0S it follows that R = 0 and
i=0
i=0
S = T. Using the r e l a t i o n F , n = F ,., + F in T, one can t r a n s f o r m the
&
s
n+2
n+1
n
solution into the f o r m S = 50 + (n3 + 6n = 12)F + 2 + (-3n 2 + 9n = 1'9)F .
Generalizing on the above technique one s e e s that
2 k Fk
k=l
u(n)F
n+Z
52
+ v(n)F
+ A
[Feb, 1964]
w h e r e u and v a r e polynomials in n of d e g r e e p and A
i s a constant independent of n.
It can be shown that the coefficients of u and
v may b e found by solving the 2p + 2 equations obtained by letting n take on

any 2p + 2 consecutive v a l u e s .
Also solved by Zvi Dresner and Marjorie
Bicknell
A CLASSICAL SOLUTION
H-16
Proposed by H. W. Gould, West Virginia University,
Morgantown, W. Va.
n x
n ~~ x
= (-1) e D (e
)e
Define the o r d i n a r y H e r m i t e polynomials by H

oo
(i)
n(x/2)ST
= X
'
Show that:
OO
yn
) H (x/2)^T-F
(ii)
= 0 ,
n=0
(iii)
^
n=0
where F
and L
a r e the n
We r e c a l l that
00
n
Hn(x/2)^L^2e"
n! n
Fibonacci and n
2 H ( t ); -7- = e
r, rv
nj
n=0
x x
'
Lucas n u m b e r s , r e s p e c t i v e l y .
For
t = 77 this r e d u c e s
2
to
2H (I = 1.
n
n\ 2 I n!
TI 4.
1 + ^5
P u t a = o
2
^
1 - \T5 ,
.
Z fx\xn _,
m
P = o . Then (a - B) 2 H 77 r~ F
p
x
2
^'
n\2/n!
n
0
9
s
(a-ar 2 )x 2
(/3=j32)x2 = 0 s i n c e a - a 2 = p - /32 = - 1 .
Similarly,
S H (f )^L
n \ 2 / n!
=e(^V
+ e (^V
= 2 e -x
n=0
See a l s o the solution in the l a s t i s s u e by Zvi D r e s n e r .
R e f e r e n c e continued from page 44.
1.
K. F . Roth, "Rational Approximations to Algebraic N u m b e r s / ' Mathematika

2 (1955) pp. 1 - 20, p 6 188.
BEGliiEHS 5 CORNER
E d i t e d by D M I T R i
THORO
California
THE EUCLIDEAN ALGORITHM I

1. INTRODUCTION
Consider the problem of finding the greatest common divisor of 34 and 144.
The factorizations 34 = 2 17, 144 = 24 32 make this a trivial problem. However , this approach Is discouraging when one deals with, say., "long" Fibonacci
numbers. Fortunately in Prop, 2 of Book VII, Euclid gave an elegant algorithm.
As usual, we shall designate the g. c, d. of s and t by (s,t).
2. THE ALGORITHM
The algorithm may be defined by the following flow chart.
A ^ B means
A replaces B, i. ee , set B = the current value of A.

M represents the remainder in the division of K by L.
I K
J L
W
->' K L
ML
/"~"^^^\Yes
M = o? \
> (I,J) = L
->f Stop j
3z:
L K
Flow Chart for Computing the G. C. D of Positive Integers I and J
53
BEGINNERS 1 CORNER
54
F o r I = 13 and J = 8,
[Feb.
the s u c c e s s i v e values of K, L ,
13
8
5
3
2
8
5
3
2
5
3
2
1
0
and M a r e :
The l a s t value of L i s the d e s i r e d g.c.d. In the following computation, (10946,

2584) = the l a s t n o n - z e r o r e m a i n d e r .
4
2584)10946
10336
4
610|2584
2440
4
" 144(610
576
4
34J144
136 4
8|34
324
2|8"
8.
0
In this d i s c u s s i o n we shall e m p h a s i z e computational
considerations.
T h e r e a r e , however, n u m e r o u s " t h e o r e t i c a l ! t applications of the Euclidean A l gorithm.
As LeVeque [1] e x p r e s s e s it,
cative n u m b e r t h e o r y . "
f?
.. . it i s the c o r n e r s t o n e of m u l t i p l i -
F o r a r e l a t e d t h e o r e m s e e Glenn Michael
[ 2 ] , this
issue.
3.
A FORTRAN PROGRAM
With an occasional glance at our flow c h a r t , it is e a s y to decipher the following F o r t r a n p r o g r a m .
( F o r t r a n i s a p r o b l e m - o r i e n t e d language
commonly
u s e d in c o n v e r s i n g with e l e c t r o n i c digital c o m p u t e r s . )
(i)
A = B m e a n s A i s r e p l a c e d by B
(ii)
The READ and PUNCH s t a t e m e n t s r e f e r to c a r d input/output.
(iii) In this context, N = K / L is an i n s t r u c t i o n to s e t N equal to [ K / L ] ,

i. e. , the g r e a t e s t Integer not exceeding K / L
fixed point quotient).
( s o m e t i m e s called an i n t e g e r o r
Thus if K = 13 and L = 3,
N will equal 4.
(iv) The symbol for multiplication i s an a s t e r i s k .
1964]
BEGINNERS' CORNER
(v)
55
A "conditional transfer' 1 is achieved by using an IF statement: if M
< 0, go to statement 3 for the next instruction; otherwise go to statement 4.

(vi) The FORMAT and END statements are technical requirements (which
may be ignored).
READ 10, I, J
10 FORMAT (315)
K = I
L = J
2 N = K/L
M = K - L *N
IF (M) 3, 3, 4
4K - L
L = M
GO TO 2
3 PUNCH 10, I, J, L
END
4.
Length of the Algorithm
A natural question arises: What is the "length" of this algorithm? I. e. ,

if s and t are given, how many divisions are required to compute (s,t)
via
the Euclidean Algorithm?

Let us designate this number by N(s,t). For convenience we may assume
s > t. Thus for n > 1, N(n + l,n) = 2; the first division yields the remainder
1, whereas the second results in a zero remainder signifying termination of
the algorithm.
(As a byproduct we see that any two consecutive integers are
relatively prime.)
In Part II we shall see how Fibonacci numbers (Fi = F ? = 1 , F., 1 = F.
1
i+i
+ F. .,) were used by Lame" to establish a remarkable result. Additional properties of N(s,t) are suggested in the following exercises.
5.
El.
EXERCISES
Note that the Euclidean Algorithm applied to the positive integers s
and t may be described by the equations
BEGINNERS* CORNER
56
8 = 1%+ r l 9
0 2= xt < s
t = rtq2 + r2s
0 < r2 < rj
= ^
0 < r3 < r2
= r
, q + r .J
n - 20
n-lTi
n
r . = r q ,n + 0
n-1
n TI+1
0 ^ r
Explain why we m u s t r e a c h a r e m a i n d e r
b e r of s t e p s .
E2.
[Feb, 1964]
< r
n-1
(r
- ) which i s z e r o In a finite n u m -
Hint: Look at the i n e q u a l i t i e s .
In E l , show that ( s , t ) = r
(the l a s t n o n - z e r o remainder),,
Use r e p e a t e d applications of P r o b l e m 1.3
Hint:
[3],
E3. (a) Verify that M = K - L * N i s the r e m a i n d e r in the division r e p r e s e n t e d by s t a t e m e n t 2 (N = K/L) of the F o r t r a n p r o g r a m .

(b) Can the F o r t r a n p r o g r a m be u s e d to compute (I, J) when I <
E4.
P r o v e that if n ^ 3 , then N(n, 3) = 1, 2, o r 3.
E5.
Suppose that n > 5 i s chosen at r a n d o m .
J?
Find the probability that
N(n,5) > 2.
E6.
P r o v e that for n > 3 ,
E7.
F o r what values of n i s 3 < N(2n - 5,n) ^ 6?
E8 8
Express
E9.
Investigate the following c o n j e c t u r e : If a - F K , then N(n,a) < K - 1.
(F
N(n + 3,n) = 2 5 3, or 4.
.., F ) as a function of n.
Can n be any positive i n t e g e r ?

E10.
number.
Investigate the following c o n j e c t u r e :
Let F ^ 2
b e any Fibonacci
Then m a x N(n, F) = 1 + m a x (n, F - 1).
REFERENCES
1.
William J. LeVeque, Topics in Number T h e o r y , Addison-Wesley, Reading,

M a s s . , 1 9 5 6 , Vol. I, Chap. 2.
2. Glenn Michael, "A New Proof for an Old P r o p e r t y , ! T Fibonacci
Quarterly,
Vol. 2 5 No. 1, F e b . 1964, p . 5 7 .

3.
D. E. Thoro ? "Divisibility I , " Fibonacci Q u a r t e r l y , Vol. 1, No. 1, F e b . 1963,

p. 5 1 .
A HEW PROOF FOB M OLD PROPERTY*

GLEN
MICHAEL
Washington State U n i v e r s i t y , P u l l m a n , Wash.
1.
INTRODUCTION
The following t h e o r e m i s c e r t a i n l y well known.

T h e o r e m : If m and n a r e .fpositive i n t e &
g e r s ,J then \(F . F / ) = F (, m , n )
m
n
F o r e x a m p l e , proofs can b e found in [ 1 , pp. 3032] and [2, pp. 148149]. In
this p a p e r we give an a l t e r n a t i v e proof which i s believed to be new.
2.
PRELIMINARY RESULTS
In addition to e l e m e n t a r y divisibility p r o p e r t i e s of i n t e g e r s , the proof d e pends on the following l e m m a s which may be found in [ 1 , pp. 10, 30 and 2 9 ] .
L e m m a 1:
F o r n ^ 0,
Y
= F
m+n
F + F F
m-1 n
m n+1 *
L e m m a 2:
F o r any n, ( F n , F n + 1 ) = 1.
L e m m a 3:
F o r n =1= 0,
3.
F o r m > 1, n > 1,
Fn J F
m n
PROOF O F THE THEOREM

we show that ( F m , F n ) = F
Let
c = (m,n)
Then e l m ,
c I n and, by L e m m a 3,
a c o m m o n d i v i s o r of F
F ).
in
and F
Also, s i n c e c = ( m , n ) ,
xi
I F
and F
and it follows that F
| Fn T h u s ,
I d w h e r e d = (F
L/ i
is
in
t h e r e e x i s t i n t e g e r s a and B such that

c = a m + bn
*This p a p e r s t e m s from a talk p r e p a r e d under the guidance of P r o f e s s o r C. T.

Long for p r e s e n t a t i o n to the Washington State University Mathematics Club.
57
A NEW PROOF FOR AN OLD PROPERTY
58
[Feb. 1964]
Since c ^ m and m and n are positive f either a < 0 or b < 0o

a ^ 0 and set k = -a 0
Suppose
Then
bn = c + km
and, by Lemma 1,
m
x
F
'
= F
bn
= F
c+km
F
c-1
+ F
km
F
c
km+1 *
Now d liF n9 d | IF m and.' .by Lemma 3,' F n IIF ,bn and F mjI F km
/ . Therefore,'
d
I F bn> d I F k m ^ " f 0 l l 0 W S f r o m ( ^ t h a t d I F k m + 1 F C B u t ( d ' F k m + 1 >'

= 1 since d j F,
and by Lemma 2, (F, , F,
1 ) = 1. Therefore, d I F .
But, as seen above, F j d. Hence, since both are positive,
(F , F ) = d = F = F ,
\ m* n .
c
(ni,n)
and the proof is complete.
REFERENCES
1. N. N. Vorob Tev, Fibonacci Numbers, Blaisdell Publishing Company, New
York and London, 1961..
2. G. H. Hardy and E. M. Wright, The Theory of Numbers, Oxford University
P r e s s , London, 1954.
SOME CORRECTIONS TO VOLUME 1, NO. 3

Page 19: On the third line from the bottom, put in > for = to read
(5 + /3 nX+1 ) >.
Page 24:
Line 5 should read, instead of "aor + 2 / 3 = 0 , "

aa + 2b = 0 .
Page 30: On line 4, change
e . n to " e " .
On line 18, change ffunitn to "limit. , ?
k PRIMER FOR THE FIBONACCI NUMBERS - PART V

VERNER
E. H O G G A T T , J R . and I. D.
RUGGLES
California
CORRECTION'
Read the l a s t displayed equation, on page 67 of P a r t IV, as
Tan { T a n " 1 i
- Tan"1 % - l }
1.
(-l)n+1 ( % ^ - )
INTRODUCTION
In Section 8 of P a r t IV, we d i s c u s s e d an a l t e r n a t i n g s e r i e s .
This t i m e
we shall lay down s o m e brief foundations of sequences and infinite s e r i e s . This

l e a d s to s o m e v e r y i n t e r e s t i n g r e s u l t s in this i s s u e and to the b r o a d topic of
g e n e r a t i n g functions in the next i s s u e and to continued fractions in the i s s u e
after that.
Many Fibonacci n u m b e r s shall a p p e a r .

2.
SEQUENCES
Definition: An o r d e r e d s e t of n u m b e r s a 1fl s a29J, a o3 ,9 ' . , a ?, i s c a l l e d an

n
infinite sequence of n u m b e r s . If t h e r e a r e but a finite n u m b e r of the a s , a~i 9 ao$
ct
,a
cL
then it is a finite sequence of n u m b e r s .

A sequence of r e a l n u m b e r s
{a } _ 1 i s said to have a r e a l n u m b e r ,
a s a l i m i t (written _^ ^ a . = a) if for e v e r y positive r e a l n u m b e r e 9 la

e for all but a finite n u m b e r of the m e m b e r s of the sequence {a }.
quence
a,
- a) <
If the s e -
{a } has a l i m i t , this l i m i t i s unique and the sequence i s s a i d to c o n -
v e r g e to this l i m i t .
If the sequence {a } fails to approach a l i m i t , then the
sequence i s s a i d to d i v e r g e .
If a
We now give e x a m p l e s of each kind.
{a } = l . l . l , - c o n v e r g e s since
a. = 1.
L
nJ
399
&
n-* n
If a = l / n , {a } = 1 , 1 / 2 , 1 / 3 , - * , l / n , c o n v e r g e s to z e r o .
If a = ( - 1 ) , {a } = 1 , - 1 , + 1 , -1,+1, d i v e r g e s by oscillation.
n
1.
s
i s , it does not a p p r o a c h any l i m i t .

If a
= n,
{a } = 1 , 2 , 3 , '
d i v e r g e s to plus infinity.
59
That
60
A PRIMER ON THE FIBONACCI SEQUENCE - PART V

Finally if a
[Feb.
= , then {a } = -^ 9 , converges to one.

n i x
n
^ s
Some limit theorems for sequences are the following:

If {a } and {b } are two sequences of real numbers with limits a and
b, respectively, then
lim ,
, u \
,i
11
v
n-ôo(a
n
+ b n ;) = a + b
lim ,
t.
u \
- b ) = a - b
n'
xa
n -*oo(
lim ,
,
n-ôo (ca ) = ca, any real c
lim
,
,
b = ab
n->oo a
n n
n l i ^ ( a n / b n ) = a/b 9 b * 0 .
3. BOUNDED MONOTONE SEQUENCES
The sequence {a } is said to be bounded if there exists a positive number, K, such that la | <K for all n > 1. If a , . ^ a , for n ^ 1 9 the
* *
I nl
n+1
n1
*
sequence {a } is said to be a monotone increasing sequence; if a a ,-.
for n ^ 1, the sequence is monotone decreasing sequence.
If a sequence is
such that it is either monotone increasing or monotone decreasing it will be

called a monotone sequence.
The following useful and important theorem is stated without proof:
Theorem 1: A bounded monotone sequence converges.
As an example, consider the sequence {(1+ l / n ) n } , this sequence is
monotone increasing and bounded above by 3.
The limit of this sequence is
well known. We will use Theorem 1 in the material to come.
4. ANOTHER IMPORTANT THEOREM

The following sufficient conditions for the convergence of an alternating
series are given below.
Theorem 2: If, for the sequence { s n } ,
1. Sj > 0,
2
(S
n-l "S n ) ( - 1 ) n
> (S
n " V l H" 1 ^ 1
>
'
f r
" 2"
1964]
61
lim g _
=
s
n~~ oo n
n+1;
then the sequence {Sn } converges to a limit s S5 such that 0 < S < S*1e
5, AN EXAMPLE OF AN APPLICATION OF THEOREM 2
For the following example a limit is known to exist by the application of
Theorem 2 of Section 4.
Let S = F / F
i-A
11
-, where {F } is the Fibonacci sequence, then S 11 ' X
11
-. .
iim
- S = x(-1) / (v F F _L1 ).; By
S
J Theorem 2 above,
9
n
' ' n n+1
n~* n
To find the limit 9 consider
n+1
F
n
11""" J.
exists.
n-1
F
n
which in terms of {S
Let the limit of S n as n tends
1
n.
nf\ is 1/S n = 1 + Sn-1
to infinity be Ss then
lim S = lim S . = S > 0. Applying the limit
theorems of Section 2, it follows that S satisfies
S = ZTT-Z
or
S2 + S - 1 =
1 +S
Thus S > 0 is given by
S =
NT5
the positive root of the quadratic equation S2 + S - 1 = 0o

6. INFINITE SERIES
If we add together the members of a sequence {a }, we get the infinite
series Sit + a2 + + a + . We now get another sequence from this
finite series.
in-
Define a sequence (S } in the following way. Let St = a t = 2 a.9 S2

n
2
n
i=l *
= 8L4 + din = 2 a.9 or in general S = a-. + a? + + a = 2 a.. This is
i=l .
called the sequence of partial sums of the infinite series.
i=l
The infinite series
62
[ Feb.
can a l s o be denoted by
A = a-! + a 2 + a 3 + + a
+
i=l
If the sequence { S n } c o n v e r g e s to a l i m i t ,
A,
S,
then the infinite s e r i e s ,
i s s a i d to converge and converge to the l i m i t S;
otherwise series A is
s a i d to d i v e r g e .
7.
1.
n-Sooa
n
2.
SPECIAL RESULTS CONCERNING SERIES
If an infinite s e r i e s
= 0.
A = a 1 + a2 + + a
c o n v e r g e s , then
This i s i m m e d i a t e s i n c e a
= S - S . .
n
n
n-1
F r o m Section 3 above, an infinite s e r i e s of positive t e r m s c o n v e r g e s
if the p a r t i a l s u m s a r e bounded above s i n c e the p a r t i a l s u m s f o r m a monotone

increasing sequence.
3 F o r the a l t e r n a t i n g s e r i e s
(-l)n+
an
such that
a n > 0, n ^ 1; a n + 1 ^ a n , n ^ 1;
ô0an= 0
n=l
then by Section 4, above, the infinite s e r i e s c o n v e r g e s ; in the t h e o r e m
Sn - I (-19 a. .
An example of an a l t e r n a t i n g s e r i e s was s e e n in P a r t IV, Section 8, of
this P r i m e r .
8.
FIBONACCI NUMBERS, LUCAS NUMBERS AND II
It i s well known and e a s i l y verified that

|
= Tan"" 1 ~ = Tan"" 1 | + Tan"" 1
Also one can verify
1964]

7 = Tan"1 \
4
63
= Tan"1 ~ + Tan"1 ~ + Tan"1 ~
or
n
_ - n . , _ - i i ' ^ - i i _
-ii
T = Tan
- + Tan
- + Tan
77 + Tan
4
We note Fibonacci and Lucas n u m b e r s h e r e , s u r e l y .
We shall h e r e easily e x -
tend t h e s e r e s u l t s in s e v e r a l ways.
In this section we shall u s e s e v e r a l new identities which a r e left a s e x e r c i s e s for the r e a d e r and will be m a r k e d with an a s t e r i s k .
* L e m m a 1: L 2
L2
- 1 = 5 F 2 -.
This i s r e a l l y a s p e c i a l c a s e of a
g e n e r a l i z a t i o n of B - 2 2 , p. 76 9 Oct. , 1963, Fibonacci Q u a r t e r l y .

L e m m a 2:
L^ = L 2 n + 2 ( - l ) n
L e m m a 3:
L 2 - 5 F 2 = 4v( - l ) n
n
n
'
*Lemma4:
L L
n n+l = W ^ " 1 ?
We now d i s c u s s
T h e o r e m 3: If
tan ^ n = l / L ,
then
T a n ( * 9 T+ ^n+9)
- 1 / F - + 1 or T a n " 1 = - i - = T a n " 1 X
2n
^2n+2> ~
' * 2n+l
F2n+1
L2n
.Tan"1
X
L
2n+2
Proof:
Tan (.0
m
L
,
:
2n
2n+2
1
+ ip
) = jy: = p
Jj
X
zn
zn+Z
2n J j 2n+2
2n+l
since
L
2n+2+L2n =
5F
2n+l
and
2nL2n+2 "
5F
2il
by L e m m a 1 above.
T h e o r e m 4: If
or
tan 6>n = 1 / F n
then Tan (<9 2n - ^ 2 n + 2 ) .'= 1
/?2n+v
64

Tan"1 ^
F
Proof:
= T a n " 1 -^- - T a n " 1

F
2n+1
2n
_
F
2n+2
te
- e
\
< 2 n " 2n + 2> " F ^ F ^
T a n
X
F
2n+2
p
2n
+
[Feb.
=
1 ~
i
F2n+1
since
F Q ô " F o =
2n+2
2n
o ,t
2n+l
nd
0 F 0 | 0 - F
2n 2n+2
0 l 1
(~l)2n+1
'
2n+l
F r o m T h e o r e m 4,
M
M
Tan ^
n=l
2"-.
n+l
Y
^
n _-^
Tan"1 J - - Tan"1
^
r
\
2n
*2n+2 /
rV
"I
= Tan
Since
..
write
M _^ 00
Tan
rp
"I
=-2 - Tan
=
= 0 by continuity of Tan
2M+2
*2M+2
x at x = 0
we
may
T h e o r e m 5:
4 =
T a n
t-J
9n4
A
^
2n+1
n=l
This i s the c e l e b r a t e d r e s u l t of D. H. L e h m e r , Nov. 1936, A m e r i c a n M a t h e -
m a t i c a l Monthly, p . 632, P r o b l e m 3801.

We note in p a s s i n g that the p a r t i a l s u m s
M
S_ =
M
iVi
> Tan - = T a n " 1 4 ~ - Tan

/ ,
F
F
r
^
2n+l
2
n=l
a r e all bounded above by Tan

1 can be applied,,
F
r
2M+2
i s monotone,,
From Theorem 3,
M
J Tan"
H.
n=l
1 = n / 4 and S
M
1
^JL- =
*2n+l
J
^
n=l
(Tan"1 - ^
\
Hn
Tan"1 ^
)
Sti+2 /
Thus T h e o r e m
1964]
65
so that
M
M
1
) Tan" -=r^
H.
2n+l
n=l
+ Tan" |
X
= 2 ^ Tan* 1 ~
+ Tan" 1
H,
-2n
^2M+2
n=l
-1
-I
-1
The limit on the left tends to Tan 1 + Tan 1/3 = Tan 2 and the right-hand
-1
side tends to this same limit and since Tan 1 / L ? M 9 -*0 9 then
Theorem 6:
I
n=l
rp
Tan
-1
T = Tan
-1 ^ 5 " 1
1 , ' - 1
-75- = $ Tan
2n
Compare with Theorem 5 in Part IV,

We shall continue this interesting discussion in the next issue.
CORRECTIONS FOR VOLUME 1, NO. 2

Page 45: In the tenth line up from the bottom, the subscripts on the Fibonacci
numbers should be reversed.
Page 47: Replace "Lamda" by "Lambda" in the title.
Page 52: In line 6, replace (Rn) with \(R n ),
In line 12, the author's name is Jekuthiel Ginsburg.
Page 55:
In problem H-18, part a, replace = by 4= .
a b
.,, / a b
Page 57: In E2, replace , with , , , , , .
Page 58: Add three dots after the 4 on the last line.
Page 60: The title "Letters to the Editor" was omitted from Fibonacci Formul a s , and, in that article, the "Correct Formula" due to the late Jekuthiel
Ginsburg is I * + 2 - 3J#n + 1*_ 2 = 3 F 3 n .
66
CORRECTIONS
Feb. 1964

Page 68: The right side of identity xix should read
I n + l " F nVl - !> .

and in identity xx, the subscript n-1 should be n-i.
The correct page number in reference 1 is 98.
Page 75: Insert three dots after /32, in line 15.
Page 80: In the last line, replace pN by p | N and p(2 3 5- -p ) by
p | (2 3 5 p n ) .
Page 81: Replace T + 1 by T
in the left side of the first displayed
equation.
Page 86: In B-12, L
= fa
], a34 = i = \T-l
instead of zero.
Page 87: Change the equations in problem B-16 to read
F2
F
F
n-1
n-1 n
Rn = 2 F - F
F2
- F nF
1
n-1 n
n+1
n-1 n
F2
F F
n
n n+1
See also solution in this issue.
Page 88:
F2
n
2F F x 1
n n+1
F2
n+1
See the last written line for notational error due to exclamation
point punctuation.
ON THE GENERAL TERM OF A RECURSIVE SEQUENCE

F R A N C I S D.
PARKER
U n i v e r s i t y of A l a s k a , C o l l e g e ,
Alaska
INTRODUCTION
It is often comforting and useful to obtain a specific formula for the general term of a recursive sequence. This paper reviews the Fibonacci and Lucas
sequences, then presents a more general method which requires the solution
of a set of linear equations. The solution may be effected by finding the inverse
of a Vandermonde matrix, and a description of this inverse is included.
THE SPECIAL CASE
As is well known, the Fibonacci sequence is completely defined by the
difference equation F(n) - F(n - 1) - F(n - 2) = 0 and the initial conditions
F(0) = 0 and F(l) = 1. If we seek a solution of the difference equation of the
form
F(n) = x ,
we obtain x - x
= 0, or x2 - x - 1 = 0.
-x
This
has two solutions; xt = (1 + \T5)/2 and x2 = (1 - '\/5)/2. Now, the theory of

homogeneous linear difference equations assures us that the most general solution is F(n) = ctxf + c 2 x n , where ct and c 2 are arbitrary constants.
(The
reader who encounters this result for the first time can verify it by substitution; the theory parallels quite nicely the theory of linear differential equations.)
The initial conditions give us two linear equations,
ct + e 2 = 0
(1)
cix1 + c 2 x 2 = 1
The solutions are ct = 1/NTS and c 2 = - 1 / ^ 5 ,

formula F(n) = l/\T5(xi
and we obtain the well-known
-x2).
The Lucas series is obtained from the same difference equation with different initial conditions. In this case, F(0) = 2, F(l) = 1, and equations (1)
become
Cj + c 2 =
ctxt
+ c2x2 = 1 " .
67
68
[Feb.
Then the general term of the Lucas sequence is

^
1 +
L(n) =
From these formulas it is possible to prove such identities as L(n) - F(n) =

2F(n - 1) and L(n) + F(n) = 2F(n + 1).
THE GENERAL CASE
We might solve all equations of the form of equations (1) by writing them
in matrix form
1
1"
2_
-C2_
(2)
x
~F(0)1
_F(1)_
and then find the inverse of the first matrix, so that

-1
F(0)
[Xl
x2J
[F(l)
The matrix
2j
is a simple case of a Vandermonde matrix, and the determination of the constants is possible if the matrix can be inverted.
Fortunately it can be easily
inverted, even if the order exceeds two.

Let us suppose that a recursion relation gives birth to a linear homogenous difference equation with constant coefficients, say,
F(n) + SLt F(n - 1) + a k F ( n - k) = 0, and that F(0) = b 0 , F(l) = b l f
F(k-l) = bk_r
1964]
69
If we seekâgain a solution of the form F(n) = x11, then we are led to the
equation
-e/ v
(3)
k-1
f (x) = x
+atx
Then assuming that the roots of (3),
+ + a, = 0
xux29a
"
are all different*, the
theory of difference equations assures us that F(n) = c t xf + c2 xf + +

c xn
k k T i i e s u ^ s e c l u e n t equations corresponding to (1), but written i n matrix
form are
1
x2
X,
k - 1
x2
k - 1
x,
(4)
k-1.
.
"k-1
or V k C = B.
Now the polynomial fj(x) = f(x)/(x - x t ) has k coefficients, and moreover fi(x2) = fi(x3) = e = f^x, ) = 0. Consequently if we form a row vector,
(written in reverse order) of these coefficients, then this row vector will be
orthogonal to every column of V, except the first.
We need now only a nor-
malizing factor, so that the scalar product of this row vector with the first column of V, is unity. Investigation shows that this scalar product (before normalizing) is fi(Xi) = V(xt)9
the fact that f'fci) = lim
XXj
the first derivative of f(x) at x = x1# Moreover,

f(x) makes this scalar product easy to calculate
X -
X1
by synthetic division.
This procedure is now continued; the coefficients of f2(x)/f?(x2) provide
-1
us with the second row of V! , and in general the coefficients of f,(x)/ff(x.)
l
th
-1
l
provide the i
row of V, .
Aparticular example makes the procedure clear. Suppose the recurrence
relation is
* When there are multiple roots, the matrix takes a different form; the inverse
for this case is not presented here.
70
ON THE GENERAL TEEMS OF A RECURSIVE SEQUENCE
[Feb.
F(n) - 3F(n - 1) - 5F(n - 2) + 15F(n - 3) + 4F(n - 4) - 12F(n - 5)

= 0,
and the initial conditions are
F(0) = 1, F(l) = 1, F(2) = 1, F(3) = 2
F(4) = 3.
The difference equation yields the polynomial
x 5 - 3x4 - 5x3 + 15x2+ 4x - 12 = 0 ,
whose roots are [-2, - 1 , 1, 2, 3] The coefficients of ft(x)
are easily found
by synthetic division,, the normalizing factor by repeated synthetic division.

2 | l - 3 - 5 + 1 5 + 4-12
-2 + 10-10-10+12
2 11-5 + 5 + 5 - 6
-2+14-38+66
1 - 7 + 19 - 33 + 60
The vector (-6, 5, 5, - 5 , 1) is orthogonal to all the columns of V5 except the
first, the normalizing factor is 1/60, and the first row of V5-1 is
(zl
JL i_ zL _L\
UO* 12* 129 12'
60 J
Synthetic division may be continued for the other roots until we obtain the desired inverse.
1
10
1
12
1
12
1
12
1
60
1
2
2
3
1
24
1
6
1
24
2
3
7
12
1
6
1
12
1
2
1
12
7
12
1
1
12 " 12
1
10
1
8
1
40
1964]
ON THE GENERAL TERMS OF A RECURSIVE SEQUENCE
Multiplying equation (4) on the left by V5 , we have
1
10
1
2
1
1
2
1
10
1
12
2
3
2
3
1
12
0
1
1"
12 60
1
1
6 " 24
1
1
6
12
1
1
12 12
1
12
'1
24
7
12
7
12
1
8
1*
20
1
12
1
J
1
1 ;
12
1
. 4 0 J H
20J
Hence the g e n e r a l t e r m i s given by

F(n) =
20 ( ~ 2 )
(_1)
12
U1)
" 12
(2)
20
(3)

P a g e 4:
Equation (2.8) should r e a d
k=o
P a g e 23:
j=o
n=0
The fifth line up from the bottom should r e a d :

D 0 ="0, D i ' = x + y , D 2 = (x + y) 2 .
P a g e 30: In Line 10, r e p l a c e
( u n + 1 - 1) by m l (u
^^ - 1)
P a g e 33: The = signs in l i n e s 10 and 11 should be r e p l a c e d by
.
= signs.
P a g e 37: The f i r s t line of the title should end in a l o w e r c a s e n m . "
71
ELEMENTARY PBOBLEMS AM SOLOTIOHS

E d i t e d by A . P .
HILLMAN
U n i v e r s i t y of S a n t a C l a r a , S a n t a C l a r a ,
California
Send all communications regarding Elementary Problems and Solutions to

Professor A. P. Hillman, Mathematics Department, University of Santa Clara,
Santa Clara, California.
We welcome any problems believed to be new in the
area of recurrent sequences as well as new approaches to existing problems,

The proposer should submit his problem with solution in legible form, preferably typed in double spacing, with name(s) and address of the proposer clearly
indicated.
Solutions to problems listed below should be submitted within two months
of publication.
B-30
Proposed by V. E. Hoggatt, Jr., San Jose State College, San Jose,
Find the millionth term of the sequence a
given that
a11 = 1,' a?l = l s9 and an+2

. 0 = an+1
, - - an
for
n ^ 1
Calif.
B - 3 1 Proposed by Douglas hind, Falls Church, Virginia
If n is even, show that the sum of 2n consecutive Fibonacci numbers is

divisible by
Fn.
J
B - 3 2 Proposed by Charles R. Wall, Texas Christian University,
Show that nL
n
Ft. Worth,
= F (mod 5).
nv
'
B - 3 3 Proposed by John A. Fuchs, University of Santa Clara, Santa Clara,
Let u ,v ,-"'9w
Texas.
California
be sequences each satisfying the second order
re-
currence formula
yn+2
where g and h are constants.
SW
+ hy
(n S
*>
Let a, b, , c be constants.
72
Show that
[Feb0 1964]
ELEMENTARY PROBLEMS AND SOLUTIONS

au n + bv n + + cwn
73
=0
is true for all positive integral values of n if it is true for n = 1 and n = 2.

B-34
Proposed by G. L. Alex under son, University, of Santa Clara, Santa Clara,
Let u
and v
California
be any two sequences satisfying the second order r e c u r -
rence formula
yn+2
syn+i
yn
where g and h are constants. Show that the sequence of products w = u v

satisfies a third-order recurrence formula
yn+3
= ay
n+2
+ b
yn+i
yn
and find as b, and c as functions of g and he

B-35
Proposed by J. L. Brown, Jr., Pennsylvania
State University,
University
Park, Pa.
Prove that
r-1
k)
k=l
for r an odd positive Integer and generalize.
B - 3 6 Proposed by Roseanna Torretto, University of Santa Clara, Santa Clara, California
The sequence 1929 5,12,29 9 70 9 is defined by ct = 1, c 2 = 29
and
^ + c for all n > 1 Prove that c r

is an integral multiple of
rt = 2c
n+2
n+1
n
5m
29 for all positive integers me
B - 3 7 . Proposed by Brother U. Alfred, St. Mary's College,
California
Given a line with a point of origin O and four positive positions A, B, C,

and D with respect to O. If the line segments OA s OB 9 OC 5 and OD c o r r e s pond respectively to four consecutive Fibonacci numbers F^ F
^ F
o F n+ Q
determine for which set(s) of Fibonacci numbers the points A S B 9 C 9 and D

are in simple harmonic ratio, i. e. ,
AB AD
BC DC "
'
74
[ Feb.
SOLUTIONS
DIFFERENCES MADE INTO PRODUCTS
B - 1 7 Proposed by Charles R. Wall, Ft. Worth, Texas
If m is an integer, prove that
n+4m+2
where F
and L
are the p
2m+1 n+2m+l
Fibonacci and Lucas numbers, respectively.
XT
XT
Solution by h D. Ruggles, San Jose State College, San Jose, California
In "Some Fibonacci Results Using Fibonacci-Type Sequences, ff Fibonacci

Quarterly 9 Vol. 1, No, 29 p. 77, it is shown that
F
q+p
q~p
= L F , for p odd.
F
p q9
If q = n + 2m -f 1 and p = 2m + 1, then this becomes the desired formula.

Also solved by Douglas Lind, Falls Church, Virginia, and the proposer.
A TRIGONOMETRIC SUM
B-18
Proposed by J. L. Brown, Jr., Pennsylvania State University, State College,
Pennsylvania.
Show that
n-1
.k cos lln-k-1
. k-~~
IT for n > 0.
F n = 2 n 1 Y (-1)^
* x | TTsin^
k=0
(It should be "for n > 1" instead of "for n > 0.?f)
Solution by the proposer
It i s well known (e. g. , I. J. Schwatt, !fAn Introduction to the Operations

with Series, n Chelsea Pub. Co. , p. 177) that
TT
cos ~
5 =
TT _
c o s
1 + ^5
4
N/5 -
_ _
1964]
75
Therefore,
1 + 4~5 0
IT
a = j
= 2 cos 1 -
b = 2 =
TT
-2sm
and
_ a -b
n-1
cos - - (-1) sin

TT
TT
cos -o + sin T^
lu
n TT
. n /
cos g - s m ^ -
TT
B
=
. (/ S)
M
n-1
^
Z CS
k=0
5 S l n (-To)
cos -= - sin f - i
>-'
n-1
= 2 11 " 1 ( - l ) k COB11"*-1 I sink - 1
k=0
as stated.
We have made use of the algebraic identity

n
- y
x - y
n-1
V
= k=0x
n-k-l
Also solved by Charles R Wall, Texas Christian University,

does not hold for n~o.
who pointed out that the
identity'
A TELESCOPING SUM
B-19
Proposed by L. Carlitz, Duke University,
Show that
Durham, N.C,
y+yi
n=l
n+2 n+3
n-1
n n+1 n+3
Solution by John 11. Avila, University of Maryland, College, Park Maryland
Our solution is similar to that by Francis D. Parker for B-9.
Let a =
a(n) = F n , b = F n + 1 , c = F n + 2 > and d = F n + 3 . Then a + b = c, b + c = d,

and the left side of the desired formula is
76
OO
OO
2
J=i
ni \xa c d
ab2d/
u
. \ abc 2 d
n=l
ab 2 cd
c - a
d - b
2(
n=l
ab 2 cd
abc d
n=l
1
:
abed
1 ^ 1
_
bc d
j-
1
2
ab c
abed
ab2c
n=l
bc2d
The l a s t s u m i s the t e l e s c o p i n g s e r i e s
\ + /_J^__i_, +
F 2 F32 F 4 J
Ft F\ F 3
\ F2 F2 F4
F3
F\
F5
whose s u m i s
FXF2F3
1 - I2 2
Also solved by the proposer,,

SUMMING GENERALIZED FIBONACCI NUMBERS
B-20
Proposed by Louis G. Brokling, Redwood City,
California
G e n e r a l i z e the well-known i d e n t i t i e s ,
(i)
F t + F2 + F8 + + F n = F n + 2 - 1
(ii)
Lj + L 2 + Lg + + L n = L n + 2 - 3
[Feb.
1964]
Solution by Charles R. Wall, Texas Christian University,
Ft. Worth, Texas
If EL = q, Hil = P f and H - = H J - + H 9,
u
>
n+2
n+1
n
so that
n
n-1
F
I H i = *J, Fl + . i = P<Fftf2-1>
1=1
i=l
i=0
77
then H = pF + HqF ,
n
^ n
n-1
<Vl-1>
= p F ^ + ql F . - (p + q) = H 1 0 - (p + q); = H , _ - H 0 .
* n+2
n+1
^ ^'
n+2 ^
^
n+2
2
This identity is also obtained from Horadam f s
quence,
Amj^cajiJV!^^
Vf
A Generalized Fibonacci Se-
Vol. 68 (1961), p. 456u
Also solved by Fern Grayson, Lockheed

Sunnyvale California and the proposer.
Missiles
and Space
Company,
EVENS AND ODDS

B . - 2 1 Proposed By L. Carlitz, Duke University,
Durham, N. C
u = | [(x+ I) 2 +(x - l) 2 *]
n
show that
u
- = u2 + 2 211 uguf - - - u2 u x
n+1
n
n-1
Solution by Robert Means, University of Michigan
on
Let Let vn = (x + 1)
- un . Then u0 = x, v0 = 1 and for n > 1 u n and
v are the terms of even and of odd degree respectively in (x + l ) 2 .
Now
un + v . = v(u + v ) ;= u2 + 2u v + v2 and equating sums of terms of
n+1
n+1
n n
n n n n
*i
&
even and of odd degree respectively we have for n ^ 0,
(a)
u n + 1 = u + v
vr2vn
78
[Feb.
= 2u 1 v 1 = 2 2 u 1 u 9 v 0 = =
n
H"~x nx
n-~x n-"^ n*^
2 u
u
* uA v n . . Since v0 = 1, the desired result is obtained by substin-~x n _ z0
u u
tuting the last expression for v in (a).
Repeated use of (b) leads to v
Also solved by Charles R. Wall, Texas Christian University and the proposer
LUCAS ANALOGUES
B-22
Proposed by Brother U. Alfred, St. Mary's College,
California
Prove the Fibonacci identity

2k*2k f
*k+k !
*k-k f
and find the analogous Lucas identity.

(Editor's Note: The Fibonacci identity here is proved by I# D. Ruggles in
"Some Fibonacci Results Using Fibonacci-Type Sequences/' this Quarterly,
Vol. 1, Issue 2, p. 77 0 ) Proofs were submitted by Douglas l i n d , Falls Church,
Virginia; V. E. Hoggatt, J r . , San Jose State College; and Charles R. Wall,
Texas Christian University, Ft. Worthy Texas. Lind and Hoggatt gave
L
2kL2j =
Lj
Lk-j -
^ " ^
as the analogous Lucas identity and Wall gave it as

L 01 L0 = L? . + 5F? . = 5 F ? , . + L? .
2k
2j
k+j
k-j
k+]
k-j
Proofs of these are left to the readers.

TELESCOPING PRODUCTS AND SUMS
B-23
Proposed by S. L. Basin, Sylvania Electronic
Systems,
Prove the identities

(i)
n /
F. 1
F , - = n x1 T+ 1 " 1
'n+1
" 1
F.
1=1 \
Mt. View, Calif.
1964 ]
79
n
n+1
(ii)
= 1+
iFt-i
lower limits m i s printed in problem
statement
i=2
1 + NT5
(iii)
= 1+
i=2
(-D x
F.F. ,
i
Solution by J. L. Brown, jr., Pennsylvania
i-l
State University,
State College,
Pa.
F , - F F0
n F., n F. + F. , n ,
*. n
l
n
2
i+l
i
i
l
i
-l
+
=
n
=
n
=
n
1
+
F.
n
F.
n+1
F F F. - F.
. n
l
i
i=l
n n-1
1 i=l
i
i=l
F3
F2
F
/F
F \
/ F
F
+ 1
n+1 _ / n+1
n_ \
I n
n-1
F
n = { Fn ~ F n - J
I V l " F n - 2 I ' " \ F> F >
(i)
(ii)
i+y
L
= 1+
i=2 V
*-* >
i=2
n
1+
i=2
F.^nF. . - F?
i+i i i
~
i
F F
t i-i
F F
using the well-known identity,

T?
i+l i - l
- F 2 = K(-11)}
i
(iii) In (ii) take the limit as n -*- and recall that

7
_ -= = r
n~*
JD
A/so solved by Dermott A. Breault, Sylvania-ARL, Waltham, Mass. ; Douglas Lind, Falls Church, Va;
Charles R. Wall, Texas Christian University, Ft. Worth, Texas? and the proposer
A CORRECTED SOLUTION
B-4
Proposed by S. L. Basin, Sylvania Electronic Systems,

and Vladimir Ivanoff, San Carlos, California
n
Fi = F
^2n
Show that
i=0
Generalize.
Mt. View,
California,
80
[ F e b . 1964]
( R e a d e r s : Can you find the e r r o r s in the p r e v i o u s l y published s o l u t i o n ? )

Solution by Joseph Erbacher, University of Santa Clara, Santa Clara, Calif., and J. L. Brown, Jr.,
Pennsylvania State University, State College,
Pennsylvania
Using the Binet f o r m u l a ,
2n+j
( a 2 ) n a 5 - (b 2 ) n b 3
a - b
(1 + a ) n a j - (1 + b ) n b j
a - b
since
b2 = b + 1
= a + 1,
when a =
1 + \r5
- 1 -
N/5
we have
r n
' 2n+j
1 / ^
a - b
i=0
1=0
i=0
n\
)
h i+ J
, ,VI
l_j
ai+j - bi+j
a - b
i=0
(K,
T h e r e f o r e , for a r b i t r a r y i n t e g r a l j ,
*-Kr
2n+j
LJ\\
i=0
If j = 0, we have the original p r o b l e m .

j,
for L u c a s n u m b e r s s i n c e L
= F
) ~i+j
The identity a l s o h o l d s , with a r b i t r a r y
. + F
..
CORRECTION TO VOLUME 1, NO. 1

See Vol. 1, No. 2 , p . 46 for c o r r e c t i o n to l a s t two r e f e r e n c e s on page 42.

2 1

Cargado por

Información del documento

Título original

Derechos de autor

Formatos disponibles

Compartir este documento

Compartir o incrustar documentos

Opciones para compartir

¿Le pareció útil este documento?

¿Este contenido es inapropiado?

Copyright:

Formatos disponibles

2 1

Cargado por

Copyright:

Formatos disponibles

THE FIBONACCI QUARTERLY

THE OFFICIAL JOURNAL OF

THE FIBONACCI ASSOCIATION

A Property of Fibonacci Numbers

Some New Fibonacci Identities

Primes Which Are Factors Of All Fibonacci Sequences . . BROTHER U. ALFRED

A Primer For The Fibonacci Numbers - Part V

On The General Term Of A Recursive Sequence

THE FIBONACCI QUARTERLY

THE FIBONACCI ASSOCIATION

WITH THE COOPERATION OF

Manuscripts intended for

Drawings should be made the same size as

Authors should keep a copy of the manuscript sent to the

A PftOPEBTY OF FIBONACCI LUMBERS

L e t A - (a l s ar, s - ) denote a (possibly finite) sequence of i n t e g e r s .

shall l e t P(A) denote the s e t of all i n t e g e r s of the f o r m

F then the r e s u l t i n g sequence

t e r m has the following r e m a r k a b l e

THE MAIN RESULTS

We f i r s t prove (B). Suppose F

We show by induction that F

We f i r s t note that the s u m of all t e r m s of F* which do not exceed

= }] < F k + 2 " F k + 1> k=l

A PROPERTY OF FIBONACCI NUMBERS

- 1 ^ P(F*) Now assume that F

t ^ 0 and consider the integer F , ? . - 1.

- 1 <j: P(F*) for some

The sum of all terms of F*

s+2t+2 + m ' w h e r e m e P"F*)" B u t m = Fs+2t+3 " Fs+2t+2 " ^ = F s + 2 t + l

be the sequence of integers defined by

Then S satisfies (C) and (D).

- 1 <f P(S*) for n > 4

We first note that

Therefore, we have (cf. Eq. (1) )

is missingfrom S*5 then

denote the sequence

i n t e g e r s w and x, P(S ) i s s a i d to have no gaps

The proof of (C)

now a consequence of the following two l e m m a s .

w > 0 and P(S ! ) h a s no gaps of length g r e a t e r than w

then t h e r e e x i s t s i such that P(S f ) has no gaps of length g r e a t e r

Proof of L e m m a 1: F i r s t note that

> (F. +2^ - F . +1n ) + n

A PROPERTY OF FIBONACCI NUMBERS

Now. let h > k + 1 and let

where pj < pj < < p.1 . Let

*' s h+2 " V l ~ s k+l

* s h+l ~ s h+2 " s k+l " e h

A PROPERTY OF FIBONACCI NUMBERS

where p " < pT2T< < p" f , we have

h > k + 1 > 5=^>s U l 0 ^ y s.

and choose n such that

A PROPERTY OF FIBONACCI NUMBERS

- m - 1), for 0 ^ m < n - p - 1

Hence, all the integers

belong to P(S f ). Since n ^ s 2

Therefore, all the integers

belong to P(S f ), i. e. 9 all the integers

so that we finally see that all the integers

A PROPERTY OF FIBONACCI NUMBERS

Since the sums

are all elements of P(Sf),