GIFT OF

Professor Robertson

BWGfNEERING

MAGIC

SQUARES AND CUBES

BY

W.

S.

ANDREWS

WITH CHAPTERS BY PAUL CARUS, L. S. FRIERSON, C. A. BROWNE, JR., AND AN INTRODUCTION BY PAUL CARUS

CHICAGO

THE OPEN COURT PUBLISHING COMPANY.

LONDON AGENTS
KEGAN PAUL, TRENCH, TRUBNER &
TQ08
CO., LTD.

COPYRIGHT BY

THE OPEN COURT

1908

PUB. CO.

To
Engineering Library

The

title

vignette

is

an ancient Tibetan magic square.

TABLE OF CONTENTS.
PAGE

Introduction.
v t
,

By Paul Carus

v
I i i

Magic Squares
General Qualities and Characteristics of Magic Squares Odd Magic Squares Even Magic Squares Construction of Even Magic Squares by De La Hire's Method Compound Magic Squares Concentric Magic Squares General Notes on the Construction of Magic Squares

18

34

44
47 54
64

Magic Cubes
Characteristics of

Magic Cubes

64
64
76

Odd Magic Cubes

Even Magic Cubes General Notes on Magic Cubes

84 89

The Franklin Squares

An

Analysis of the Franklin Squares.

By Paul Carus

96
113 113 120 122 125

Reflections on

Magic Squares. The Order of Figures

By Paul Carus
-,

Magic Squares in Symbols The Magic Square in China

The Jaina Square

Mathematical Study of Magic Squares.

By

L. S. Frierson

129

A New Analysis A Study of the Possible Number

Notes on Number
Squares
Series

129
of Variations in
in

Used

Magic Squares.. 140 the Construction of Magic

.

148

Magic Squares and Pythagorean Numbers. By C. A. Browne Mr. Browne's Square and lusus numerorum. By Paul Carus

156 168
173

Some Curious Magic Squares and Combinations

Notes on Various Constructive Plans by which Magic Squares
Classified

May

be
185

I*

'

The Mathematical Value

of

Magic Squares

194

868516
.)o

*2

/t-t*-*^

VH^Z&tSt
ct^f***.

0-4. 4KXc.

Jrt~

INTRODUCTION.
r

I
-*

HE

peculiar interest of magic squares and

all

lusus

numerorum
which by a

in general lies in the fact that they possess the

charm of mys-

tery.

They

appear to betray

some hidden

intelligence

preconceived plan produces the impression of intentional design, a

phenomenon which

finds

its

close analogue in nature.

Although magic squares have no immediate practical use, they have always exercised a great influence upon thinking people. It seems to me that they contain a lesson of great value in being a
palpable instance of the
a clear light

symmetry of mathematics, throwing thereby

that pervades the universe wherever

upon the order

we

turn, in the infinitesimally small interrelations of

atoms as well

as in the immeasurable

which, although of a different kind and

domain of the starry heavens, an order still more intricate, is also

life,

traceable in the development of organized

and even

in

the

complex domain of human

action.

Magic squares
this evidence

are a visible instance of the intrinsic

harmony

of the laws of number, and

we

are thrilled with joy at beholding

which

reflects the glorious

symmetry of the cosmic

order.

Pythagoras says that number is the origin of all things, and certainly the law of number is the key that unlocks the secrets of the universe. But the law of number possesses an immanent order,

which
tance

is

at first sight mystifying, but

we

easily

understand

it

on a more intimate acquainto be intrinsically necessary; and this

instances of the intrinsic

law of number explains the wondrous consistency of the laws of

nature.

Magic squares are conspicuous

VI

INTRODUCTION.
will serve as

harmony of number, and so they

cosmic order that dominates
all

an interpreter of the

existence.
intellectual play that illustrates the

Magic

squares are a

mere

nature of mathematics, and, incidentally, the nature of existence

dominated by mathematical regularity. harmony of mathematics as well as the

of the cosmos.

They

illustrate the intrinsic

intrinsic

harmony of

the laws

In arithmetic
counting; in

we

create a universe of figures by the process of

lines
;

geometry we create another universe by drawing

in the abstract field of imagination, laying in algebra

down
more

definite directions

we produce magnitudes
letters.

of a

still

abstract nature, ex-

pressed by

eral conditions in

In all these cases the first step producing the genwhich we move, lays down the rule to which all further steps are subject, and so every one of these universes is dominated by a consistency, producing a wonderful symmetry, which
in the

cosmic world has been called by Pythagoras "the harmony of

the spheres."

There

is

no science that teaches the harmonies of nature more and the magic squares are
like

clearly than mathematics,

a magic

mirror which

reflects

a ray of the symmetry of the divine

norm

immanent

in all things, in the

not less than in the mysterious depths of the

immeasurable immensity of the cosmos human mind.

PAUL CARUS.

-1

te

MAGIC SQUARES.
AGIC
itful

squares are of themselves only mathematical curios,

but they involve principles whose unfolding should lead the

mind

to a higher conception of the wonderful laws

symphony and order which govern the science of numbers. ot record of a magic square io found
subject liaj
iiuiL ui

leas 3tudicd

and developed

and compre"T^rpose to present some gen methods for constructing magic squares hensive tich he believes to be original, and also to B^ieflv review commonly known
It is the writer's
itraefe

THE GENERAL QUALITIES

flND CHARACTERISTICS

OF

MAGIC SQI\KS.
magic square consists of a JfrieV of numbers arranged in quadratic form so that the sumJof eachVertical, horizontal and
corner diagonal column
is

the^ame amounBL These

squares can

be

made with

either

an odd

yan

even number^tf

cells,

but as odd

squares are constructed byJmethods which differwom those that govern the formation of Jfren squares, the two class%.will be considered under separate J^adings.

)DD MAGIC SQUARES.

it is not only requisite that the sum "3^ squj^es the same amount, but also that the sum of columns shall

In these

all

MAGIC SQUARES.

E5
j&'4&
3
is

OF

ODD NUMBERS.
i

A"

*.*i

*U &*&"s<Bare

shown

in Fig.

covers the smallest ag-

gregation of numbers that

carJable of

magic square arrangement,

and

it is

also the only possible

relatively
will

arrangement of nine different numbers, to each other, which fulfills the required conditions. It

be seen that the

sum

of each of the three vertical, the three

horizontal,
15,

and the two corner diagonal columns in this square is making in all eight columns having that total: also that the
of any

sum

two opposite numbers

therefore Q poyioot

is

10,

which

is

twice the center

number.
le

Itmie

ocfuai'L uf 3 )(
is

^.

next largest odd magic square

that of

5X5,

and there

are a great

many

different arrangements of twenty-five numbers,

/7

Totals

15.

MAGIC SQUARES.
angles to that which was last considered, so that
cylinder with the extreme right and
left
it

forms a

vertical

hand columns adjacent

assist

to

each other.

An

understanding of this simple conception will

to follow the-trew

the

are to be
right or left

methods of building odd magic squares of thooo mothodo being,' based on a described/ hand diagonal formation.
all'

Referring to Fig.

2, it will

be seen that the square

is

started

by writing unity

in the center cell of the

numbers proceeding diagonally therefrom

upper row, the consecutive in a right hand direction.

Using the conception of a horizontal

cylinder, 2 will be located in the

cell to

lower row, followed by 3 in the next upper

the right.

Here

the formation of the vertical cylinder being conceived the next upper

square will be where 4

here blocked by
I

is

written, then 5

further progress being

cell

which already occupies the next upper

in

diagonal order.

When
in this case

a block thus occurs in the regular spacing (which will

be at every fifth

number

in a 5

5 square) the next

number must
last filled,

be written in the
written

cell vertically

below the one

so that 6

is

in the cell

below

order

is

then continued in cells

is

5, and the right hand diagonal occupied by 7 and 8. Here the

horizontal cylinder

imagined, showing the location of 9, then the

;

conception of the vertical cylinder will indicate the location of 10

further regular progression being here once
so ii
is

more blocked by

6,

written under 10 and the diagonal order continued to 15.

A
n,

mental picture of the combination of vertical and horizontal cyl-

inders will here

so 16
is

show

that further diagonal progress

is

blocked by

written under 15.

The

vertical cylinder will then indi-

cate the cell in

will

which 17 must be
cell

located,

and the horizontal cylinder

reached and the

show the next

1

diagonally upwards to the right to be occuuntil the final

pied by

8,

and so on

number 25

is

square completed.
Fig. 3 illustrates the development of a 7

7 square constructed
is

according to the preceding method,

and the ctudont

advised to

follow the sequence of the numbers to impress the rule on his

ory.

mem-

variation of the last

method

is

shown

in Fig. 4, illustrating

MAGIC SQUARES.
another

7X7

square.

In this example

is

placed in the next

cell

horizontally to the right of the center

cell,

and the consecutive

numbers proceed diagonally upward therefrom, as before, in a right hand direction until a block occurs. The next number is then
written in the second
filled

cell

horizontally to the right of the last cell

(instead of the cell below as in previous examples) and the

is

upward diagonal order

30 33 45 3S
/

resumed

until

the next block

occurs.

/O

26
2 7 29

7
6
26

2S
/3 /s
2/ 23

36 4*

33 42 44
/2

22

2O

Fig.

3-

MAGIC SQUARES.
previously referred
to.

To

chess-players the knight's

move

will

require no comment, but for those who are not familiar with game it may be explained as a move of two trqunrag straight

this

for-

ward

in

any

direction

and one aquae to

of 5

either right or left.

The magic square

placing
i

in the center cell of the in its

move employed
one
cell to

by and the knight's upper row, construction will be two cells upward and

X5

illustrated in Fig. 5 is started

the right.

in the

Using the idea of the horizontal cylinder 2 must be written second jfape from the bottom, as shown, and then 3 in the
from the
top.

second &t

Now

conceiving a combination of the

horizontal and vertical cylinders, the next

move

will locate

in the

extreme lower

left

hand corner, and then

5 in the
<***,

middle row.
is

We

now
5,

find that the next

move

is

blocked by

so 6

written below

last

and the knight's moves are then continued, and so on until the number, 25, is written in the middle cell of the lower fae; and
In

)fcthe square is thus completed.

common
it

described,
1

with the odd magic squares which were previously will be found that in this square the sum of each of

the five horizontal, the five perpendicular, and the

two corner diagonal columns is 65, also that the sum of any two numbers that are ' gcomctricallyfequidistant from the center is 26, or twice the number
in the center cell, thus filling

all_thegeaaafr qualifications of a
will be noted

perfect square.

In addition, however, to these characteristics

that each spiral

it

figures around the horizontal and vertical cylinders traced either right handed or left handed also amounts In the vertical cylinder, there are five right hand, and five to 65.
left

row of

hand

spirals,

umns

across the square, leaving eight

two of which form the two corner diagonal colnew combinations. Tho oamc

Counting therefore

five horizontal

columns, five vertical col-

umns, two corner diagonal columns, and ciiitoen right and left hand spiral columns, there will bo foand in all twenty-^gfat columns
each of which will

sum up

to 65,

whereas

in the

5X5

square shown

MAGIC SQUARES.
in Fig. 2 there will

be found only twcfoc columns that will amount

to that number.

This method of construction

is

subject to a

number of

variations.
left

For example, the knight's move may be upwards and to the

hand instead of
to the right, or
it

may

be made downwards and

in other directions.

either to the right or left hand,

and also
in

There

are in fact eight different

ways
cell

which the knight's move may

upper

be started from the center

in the

^t.

Six of these

moves are indicated by figure 2's each of these moves if continued

ruiiukir
tjcj^tx*^

in different cells of Fig. 6, in its

and
by

own

direction, varied

breaks as before described, will produce a different but

pprfppt square.

The remaining two

possible knight's moves, indi-

cated by cyphers, will not produce perXect

Fig. 6.

Fig.

7.

It

may

here be desirable to explain another method for locating

numbers

which some may prefer to that which involves the conception of the double cylinder. This method consists in constructing parts of auxiliary squares around two or more
in their

proper

cells

sides of the

main square, and temporarily writing the numbers

in

the cells of these auxiliary squares

ries

when

their regular placing car-

them outside the

limits of the

main square.

The temporary

location of these

numbers

in the cells of the auxiliary squares will

then indicate into which

cells of the

main square they must be per-

manently transferred. Fig. 7 shows a 5

main square with parts of three auxiliary

MAGIC SQUARES.
squares, and the
5-

main square

will

be built up in the same way as

Starting with

in the center of the top

UH^

the

first

knight's

move

of two cells upwards and one to the right takes 2 across the
cell

top margin of the main square into the second

of the second

from the bottom

transferred to the

in

one of the auxiliary squares, so 2 must be

relative position in the

same

main square.

Start-

ing again from 2 in the main square, the next

move

places 3 within

main square, but 4 goes out of it of an auxiliary s4uare, from which

the

into the lower left hand corner

it

must be transferred

to the

same

main square, and so on throughout. The method last described and also the conception of the double
location in the

cylinders

little

be considered simply as aids to the beginner.* With '/tf^ practice the steidont will be able to select the proper cells in 7

may

the square as fast as the figures can be written therein.

^
lines of con-

Having thus explained tii&Mr

struction, the general principles

"p^rifir

Tfpi]

MINI

1-

governing the development of odd magic squares by these methods may now be formulated. i. The center cell in the square must always contain the middle

number of
/T\ \S ^,
is
|

the series of

numbers used, /

i.

e.,

a number which
last

equal to one half the sum of the

first

and

numbers of

the series^ <rf2.

7^

*^_

-f-

No p^^at
center
cell,

but

magic square can therefore be started from its it may be started from any cell other than
which
will

the center one.

3.

With

certain specific exceptions

be referred to

later on,

right or

left

odd magic squares may be constructed by either hand diagonal sequence, or by a number of so-

called knight's moves, varied in all cases

4.

by periodical and from normal spacing. The directions and dimensions of these departures from
well defined departures

normal spacing, or "break-moves," as they may^betermed,

are governed by the relative spacing of cells occupied by the
first

C~~y

and

last

numbers of the

series,

and may be

deter-

mined as follows:

MAGIC SQUARES.

if*

nv

RULE: Place
in the cell

the

first

number of

(excepting the center one) and the/last

the series in any desired cell number of the series

cell

which
first

is

gGomotricallyroQnooito to the

con-

t/~

taining the
cell that

number.

The

relative spacing

between the

contains the last

first

number

of the series and the cell

that contains the

number

of the series must then be

repeated whenever a block occurs in the regular progression.

EXAMPLES.

Using a blank square of

cell

5,

imaV

be vfritten

in the

middle

opposite being 25 must be written therein. I will inthe^lpwertase, "thereforeTielocated four cells above in the middle vertical column,
the middle cell

of the upper jaacr

The goomctrkally

cell to this

or what

is

the

same

thing,

and

easier to follow, one cell

below

25.
first

When, therefore, a square of 5 X 5 is commenced with the number in the middle cell of the upper fae, the break=:move
gdiiiiTjrrbe o

will

,-/,

MAGIC SQUARES.
will naturally

vary with the method of advance, but the relative

spacing of the moves themselves will remain unchanged.

NOTE The foregoing break=moves were

:

previously described in
I, 2, 3, 4,

several specific examples (See Figs.

and 5) and

the reader will

eral rule.

now

observe

how

they agree with the gen-

in
I

Once more using a blank square of 5 X 5, the upper left hand corner and 25 in the lower
will then

may be

written

right

hand corner.
left

occupy a position four cells removed from 25 in a

hand upward diagonal, or what is the same thing and easier to follow, the next cell in a right hand downward diagonal. This will
therefore be the breakrmove whenever a block occurs in the regular

Fig 10 shows the break moves which occur when a knight's move of two cells to the right and one cell upwards is used
spacing.
for the regular advance.

As
left in

a final example

we

will write

in the second cell

from the
placing

the upper line of a 5

5 square, which

calls for the

2,

IO

MAGIC SQUARES.

such as right hand diagonal sequence, upwards or downwards, left hand diagonal sequence upwards or downwards, or a number of knight's moves in various directions. There are four possible moves

from each
each
cell

cell in

diagonal sequence, and eight possible moves from

by the knight's move.

Some

of these moves will produce

ggrfTttfr

magic squares, but there will be found

many

exceptions

which can be shown most readily by diagrams. Fig. 12 is a 5 X 5 square in which the pointed arrow heads
dicate the directions of diagonal sequence by

in-

which perfect squares

may

be constructed, while the blunt arrow heads show the directions

MAGIC SQUARES.
Fig.
is
1

II

6 shows a similar square

in

used for regular advance.

The break ^move

which an elongated knight's move is necessarily the same

in

each example.
s

(See Fig. 10.)

2^-2-

-*,

/,

Z-/

12
Figs. 21, 22,
I

MAGIC SQUARES.
and 23 illustrate three 5X5 squares, each having upper right hand corner and 25 in the lower left hand

in the

/O 22

9
24

20
2/

26

/6

Fig. 21.

MACK
equivalent to a right hand

SQUARES.

downward diagonal sequence wherein

filled.

alternate cells are consecutively

There are

in fact

almost innumerable combinations of moves

by which -&Mfi04 odd magic squares may be constructed.

63
6s
/fl

7
6-6

7*

7 66
'7

J/

77

4%

76

4'

6s

Totals

= 369.

7*
zs

39 26
' 4,0

60 3$

36

7*

7"

JS
Fi

s/
f.

24-

39

20

77 7*
6* 6A
/6

49 44
S3

40

4* J/

2.6
22,

7
Totals

6s 6* 46 4/ 36

= 369.

7"
66

A3

41

23

7* 7'
23
/a
Fig.

SZ JS 33
6? 61

43

foregoing method for building odd magic squares by a

continuous process, imic hiring- the regular spacing of consecutive

numbers varied by
struction have been

different well defined break-smoves.i

writer, but other

methods of con-

known

for

many

years.

One

of the most interesting of these sldur methods involves

the use of

two or more primary squares, the sums of numbers

in

MAGIC SQUARES.
similarly located cells of

which constitute the correct numbers for

transfer into the corresponding cells of the

to be constructed therefrom.

magic square that

is

This method has been ascribed primarily to

been more recently improved by Prof. Scheffler.
It

De

la

Hire but has

may

be simply illustrated by the construction of a few

Figs. 26 and 27
i

5X5
two

squares as examples.

squares in which the numbers

to 5

show two simple primary are so arranged that like numin the

bers occur once and only once in similarly placed cells in the

squares

also that pairs of unlike

in

numbers are not repeated

cells.

same order

any similarly placed

cell in

Thus, 5 occupies the ex-

few of each square, but this combination does not occur in any of the other cells. So also in Fig. 27 4 occupies the extreme right hand cell in the upper line, and in Fig.
treme right hand
the lower

26

this cell contains 2.

No

other

cell,

however,

in

Fig. 27 that

contains 4 corresponds in position with a cellin^26 Leaving the numbers in Fig. 26 unaltered, the numbers in Fig. 27

that contains 2.

must now be changed to

ducing the4*ey square'
bers of the
^nttsttfi*

their respective
in Fig. 28.

key numbers, thus pro-

shown

By adding

the

cell

num-

square Fig. 26 to the corresponding

cell

numbers

numbers,.

i,

2,
5,

3,

4,

5.

numbers,

o,

10, 15, 20.

MAGIC SQUARES.
It is

^*"^^

W-

28 may be pensed with by mentally substituting the kt* numbers for the
in Fig.

evident that the JOT? square

shown

dis-

l6

MAGIC SQUARES.
x

puimapv square Fig. 34 to the

bear

square Fig. 36, the latter

numbers occupying similar cells in being derived from the primary cquarc

MAGIC SQUARES.
written in heavy figures, have the correct summation.
in these

The numbers
must

two columns must therefore be

left

as they are, but the locain light figures,

tion of all the other

numbers, which are written

simple method for effecting this change consists in substituting for each number the complement between it and 17. Thus, the complement between 2 and 17 is 15, so 15 MMM* be written be changed.
in the place of 2,

and so on throughout.

All of the light figure

/z

Fig. 44.

Fig. 45..

numbers being thus changed, the square shown in Fig 42.

result will be the ponfatt

magic

***"

arrangement of figures may be attained by leaving the light figure numbers in their original positions as shown in Fig. 43, and changing the heavy figure numbers in the two corner diagonal columns to their respective complements with 17.
relative
It will

The same

be seen that this

is

only a reversal of the order of the figures

JS
JO

32

7
'7 2/ 22 20 /s

/2 26

29
33
36

J/

20

MAGIC SQUARES.
There are many other arrangements of sixteen numbers which

will fulfil the required conditions

but the examples given will

suffice

to illustrate the principles of this square.

The next even magic square many variations is shown in Fig.

is

that of 6

X 6,

and one of

its

46.

An

analysis of this square

2S

'7 2/ 22
/2 26

29

Fig. 48.

with the
to its

aid'

of geometrical diagrams will point the

way

not only

own

reconstruction but also to an easy

method

for building

6X6 squares
Fig. 47

shows a 6

6 square

in

which

all

the

numbers from

'9

Fig. 49.

to 36 are written in arithmetical sequence,

and the twelve numbers

in

in the

order,

two corner diagonal columns will be found all other numbers requiring rearrangement.
numbers
in the diagonal

magic square

Leaving there-

fore the

columns unchanged, the next step

will be to write in the places of the other numbers their complements

In this square with 37, making the square shown in Fig. 48. out of the total of twenty-four numbers (written in heavy figures)

MAGIC SQUARES.
thirty-six

21

numbers,

will

be found in magic square order, twelve

numbers (written

in light figures)

being

still

incorrectly located.

Finally, the respective positions of these twelve

numbers being

re-

versed in pairs, the magic square given in Fig. 46 will be produced.

Fig.

50 shows the geometrical diagrams of this square,

first

A
ir*

being a diagram of the

fifth lines,

and sixth

lines,

of the second and

and

of the third and fourth lines.

The

striking

regularity

of these

diagrams points to the imperfection- of the

square which they represent, in which, although the

sum

of each

xe

^c

of the two corner diagonal, the six horizontal, and the six perpendicular columns

in, yet only in the two diagonal columns does the sum of any two numbers which occupy/gcomctrioQJ-ljr oppooite aeUe,^
is

'^amount

to 37, or the to

sumof

the firstandTasTnumbers of the series.

irregularities,

Owing

their

pronounced

these diagrams convey

Fig. 50.

but

little

meaning, and in order to analyze their value for further

constructive

work

it

will be necessary to

go a step backwards and

make diagrams of the intermediate square Fig. 48. These diagrams are shown in Fig. 49, and the twelve numbers therein which must
be transposed (as already referred to) are marked by small circles

around

dots,

each pair of numbers to be transposed in position

line.

being connected by a dotted

The numbers

in the

two corner

diagonal columns which were permanently located from the be-

ginning are marked with small

circles.

We

have here correct geometrical figures with

definite

and well

defined irregularities.

The

series of geometrical figures

all

shown
6

in
r~~"^"

A, B, and

remain unchanged in shape for

variations of

X6

f^

squares, but by modifying the irregularities

we may

readily obtain

22

MAGIC SQUARES.

the data for building a large

all

number of

different

qtmres,

showing, however, the

A
of 6

series

same general characteristics as Fig. 46. of these diagrams, with some modifications of their
given in Fig. 51, and in order to build a variety
squares therefrom
it is

irregularities, is

X 6 magic

only necessary to select three

X X X X

/f

Fig. 51 (First Part).

diagrams

in the

order A, B, and C, which have each a different form

of irregularity,

and after numbering them

in arithmetical

sequence

from

to 36, as

shown

in Fig. 49,

copy the

numbers

in

diagrammatic

order into the

It

cells

of a 6

X6

square.
cells in the

must be remembered that the

corner diagonal

MAGIC SQUARES.
columns of
the

Wen

magic squares may be correctly

filled

by writing

numbers

in arithmetical order according to the rule previously

given, so in beginning any

to first write the

new even square

it

will

be found helpful
will then serve

numbers

in these columns,

and they

as guides in the further development of the square.

X X X X

2/

Fig. 51

(Second Part).

Taking
it

will

example the be seen from Fig. 49 that

for
I

6X6
it

magic square shown in Fig. 46, is constructed from the diagrams

Fig. 46
35,

marked

9 and 17

in Fig. 51.

with diagram A, Fig. 49,

in

34 unbroken order; then the diagram shows that 33 and 3 must be

Comparing the first IUK of the sequence of numbers is I,

MAGIC SQUARES.
transposed, so 3
is

written next (instead of 33) then 32 and 6 in

last Jine
-

unbroken order.
,

In the

**

of this square

31 comes

first,

then, seeing that 5

;

is

written instead of 5
is

then 4

(still using diagram and 2 must be transposed, then as 3 and 33 must be trans-

posed, 33

written instead of

3,

5 instead of 2,

and the &6T1?

finished with 36.

Diagram B gives the development of the second

TABLE SHOWING 128 CHANGES WHICH MAY BE RUNG ON THE TWENTY-FOUR DIAGRAMS IN FIG. 51.

A
i,

B
9
10
ii
17,

C \^
"
18,

2,

3 or 4

IQ or
" "

" "

" "

20=16 "
" "
"
"

12
5, 6,

"

7 or 8

13 14 15 16

21, 22, 23 or " " "

= 16 changes == 16 = 16 " 24= 16 = 16 = 128

=16
"

Total changes

EXAMPLES.

3S
28

J3

2S
20
/a

'7 /S
/3

30 26
J/

/Q

MAGIC SQUARES.
of presenting one of these squares ready

made and analyzing

of this
size.

it,

we

will

now

use the information which has been offered by previous

examples

in the construction of a

new square

Referring to Fig. 45, the regular geometrical diagrams of the

4X4

square naturally suggest that an expansion of the same

may

be utilized to construct an 8

8 square.

This expanded diagram

Fig. 52.

is

accordingly shown in Fig. 52, and in Fig. 53

is

we have
in

the magic

square that
order.

produced by copying the

numbers

diagrammatic

Fig. 53-

As might be
rfirtdrirtiV,

anticipated, this square

it

is

p effect

and the ease with which

has been constructed points

to the simplicity of the

method employed.
in Fig. 53
is,

The magic square shown

however, only one of a

26
multitude of 8
characteristics

MAGIC SQUARES.

8 squares,

all

of which have the same general

facility

and may be constructed with equal

from

X
><
Fig. 54-

X
XJ
ft

25-

X
J3
Fig. 55-

various nogulw diagrams that can be readily derived from transpositions of Fig. 52.

Five of these variations are illustrated in Fig.

MAGIC SQUARES.
54,

27

which also show the transpositions by which they are formed from the original diagrams. To construct a poafoot magic square from either of these variations it is only necessary to make four
copies of the one selected, annex the

numbers

to

64

in arithmetical

60
/o

62,

S3 SZ

/s

2' 20 22 42
-JJ

29

7
J/ 32

Totals

= 260.

zs 26
23 43

7
J'J-

49

Fig. 56.

order as before explained, and then copy the numbers in diagrammatic sequence into the
It will

cells

of an 8

X8

square.

be noted in the construction of the

4X4

and

8X8

ft,

60

28

MAGIC SQUARES.

7
JZ
20 22 JO
6-6

zf
32 J/

JJ
Totals

= 260.

26
26

/J /z
S<f

Fig. 58.

:x

Fig- 59-

ss
/o
/J S2

/s 49 42 '7

V
23 3S
JJ

JZ
Totals

= 260.

26 JO
ZA

46 /f

/A SO
60

Fig. 60.

MAGIC SQUARES.

shown

annexed examples. Fig. 55 illustrates two different forms from which the magic square Fig. 56 is constructed. Fig. 57
in the

shows three

different

forms which are used

in connection

with the

square in Fig. 58, and in a similar manner Figs. 59 and 60 show

four different diagrams and the square derived therefrom.

The

Fig. 61.

foregoing examples are sufficient to illustrate the immense number

of different

8X8

magic squares that may be constructed by the

to the

aid of various diagrams.

We

now come

magic square of 10
examples,
it

10,

and applying
be easy to ex-

the comparative method

to the last

will

MAGIC SQUARES.
pand the three diagrams of the
diagrams
that are

6X6

square (Fig. 49) into

five

required

for the

construction of a series of

99
o
8f>

96

92

ss '7 SJ
7*

77
66
Sff

26

2<f

7/
J-/

69

#6.

Totals

= 505.

E
2f>

36
29

7s
/S"

7 s 72
S? /a $9

a
Fig. 62.

/oo

10

10 squares.

These

five

diagrams are shown

is

in Fig. 61,

and
the

in Fig.

62

we have

the magic square which

made by copying

Fig. 63 (First part).

numbers from
10

to 100 in diagrammatic order into the cells of a

10 square.

It will

be unnecessary to proceed further with the construction

MAGIC SQUARES.
of other 10

10 squares, for the reader will recognize the striking

resemblance between the diagrams of the

6X6

and the 10

10

squares, especially in connection with their respective irregularities.

Fig. 63

(Second part).

'42

/Z
23

/S
J>
/o6

9S
62,

SO

*7

/6

77 7* 7
6s SS

7/

Totals

= 870

7Z 7* 7 s

7*
32.

S3

JZ
'Z

/J/

/33

7
Fig. 64

It will

also be seen that the

same methods which were used

for

32
varying the 6

MAGIC SQUARES.

X6

diagrams, are equally applicable to the 10

10

diagrams, so that an almost infinite variety of changes may be rung on them, from which a corresponding number of 10 X 10 squares may be derived, each of which will be different but will resemble
the series of 6
perfections.

X6

squares in their curious and characteristic im-

Fig. 65 (First Part).

l^even magic
d
it

is

worthy of
side ot at*

is
^^Mi^^

an even numberHtescpiare can be made P

U Ul

~ U.-.MJ

MAGIC SQUARES.

33

Fig. 65

(Second Part).

/35
/SO
/6s

/OO
/S
/

/a,

l_
2,3

7s 20
/6A Aff

'2L

'7*
/A'J

67
/si

At
st

40

Aff

/fS

ss /AS

66
//J

/zf
/*_

7*
//O

ft

ft

ss /OS
/t>2

32,

90

/OS
94.

/O/

*7

9S
X/3

#>
//a

/os~ /oA

/If
/It

7*

77
/A A
62,

r_

A6

6*

*/#

A0
2S '7

26
/Sf

7*

'?*

Z/

2.1,

'77
1 94

Fig. 66.

34
square in Fig. 64
is

MAGIC SQUARES.
is

derived.
that

The geometrical design

in Fig. 52 for the

of these
square,

diagrams and it is manifest that

the

same as

shown

8X8

all

the variations that were 12

made

in the

8X8

diagrams are also possible in the

12 diagrams, besides an

immense number of additional changes which are allowed by the

increased size of the square.

In Fig. 65

we have

a series of diagrams illustrating the de-

velopment of the 14

14 magic square shown in Fig. 66.

These

diagrams being plainly derived from the diagrams of the 6

10

X6

and
it is

10 squares, no explanation of them will be required, and

evident that the diagrammatic method

the construction of
It
all sizes

may

be readily applied to

of even magic squares.

illustrate

will

be nwLCTl that the foregoing diagrams

in

graphic manner the interesting results attained by the harmonious

association of figures,
infinite variety

and they also

clearly demonstrate the almost

of possible combinations.

MAGIC SQUARES.
3.

35
all

Construct another

4X4

square, having

numbers

in the

same positions

relatively to each other as in the last square,

all

but reversing the direction of

horizontal and perpendicular

columns (Fig. 69).

4.

Form
in

the

jgsf.

square Fig. 70 from Fig. 69 by substituting

for
piii^iic

lyy numbers
ft>&&-

numbers, and then add the numbers

Ui*e4*y^ye
shown

to similarly located

numbers

in the

primary

3bj

square Fig. 68.

The

result will be the perfect square of

4X4

in Fig. 72.

the 4np square Fig. 71 from the primary square 68 and adding the numbers therein to similarly located numbers Fig. in the primary square Fig. 69, the same magic square of 4 X 4 will be produced, but with all horizontal and perpendicular columns re-

By making

versed in direction as shown in Fig. 73.

NUMBERS
I

NUMRERS

2
3

4
8
12

MAGIC SQUARES.

The

cells

of two

6X6

squares

may

be respectively

filled

with

and Js&y numbers by analyzing the contents of each cell in Commencing at the left hand cell in the upper row, we
note that this
cell

contains

i.

In order to produce this number by

it

the addition of a
"O

iiic number to a -kw number

selected

is

evident that

and

must be

and written

into their respective cells.

The second number Jn the top row of Fig. 46 being 35, the kej^ number 30 must be written in the second cell of the 4*y square and the JMMJMT number 5 in the second cell of the pgimc square, and
on throughout Figs. 74 and 75.
all

the cells, the finished squares being

shown

in

-rv^t

Another ptioae square may now be derived from the k*y square Fig. 74 by writing into the various cells of the former the

MAGIC SQUARES.
to the similarly placed cell

37

numbers

in Fig. 75.

Having thus
its

in-

versely traced the development of the

magic square from

A and

primary and 4jeysquares,

it

will

be useful to note some of the general

characteristics of even primary squares,

which govern

their construction, as these rules will be

and also to study the rules found in-

structive in assisting the sfr*4gt to

work out an almost endless

square shown in Fig. 75,

variety of even magic squares of

I.

all

dimensions.
it

Referring to the 6
the numbers

X 6 A primary

will.be noted that the

I

two corner diagonal columns contain

order, starting respectively

to

m arithmetical

from the upper and lower left hand corner cells, and that the diagonal columns of the B primary square in Fig. 76 also contain the same numbers in arithmetical order but starting

MAGIC SQUARES.
4.

The sum

of every column in a 6
it

6 ke*F sqt ke/' square must be 90,

and under these conditions

follows that the

is

sum

of every

formed by the combination of a primary square with a k&f- square must be

column of a

6X6

magic square which

in (21+90=111).
5.

~ K><rtin

With the necessary changes

good
for all
sizes

numbers the above

rules hold

of

and

primary squares and

We

may now proceed

to

show how a

variety of 6

6 magic
in

squares can be produced by different combinations of numbers

/

MAGIC SQUARES.
Fig. 78

39

of these 6

shows the sequence of numbers in the diagonal columns X 6 A primary squares, and as this arrangement cannot
series, the

be changed in this
Fig. 77 must be

various horizontal columns or rows in

selected accordingly.

The
4.

small letters at the right

No.
5.

No.

i.

No.

2.

No.
b

3.

No.

No.

6.

f
c

d
e

a
Fig. 79-

of Fig. 78 indicate the different horizontal columns that

for the respective lines in the square
in Fig.
;

may

be used

thus either

a, b,

or c column or / for the

77

may
c,

be used for the

first

and sixth

lines, a, e,

second and
neither b,

fifth,

and

c,

d,

or e for the third and fourth lines, but

fifth lines,

or d can be used in the second or

and so

forth.

Six different combinations of columns are given in Fig, 79,

from which twelve different

structed.

6X6
I

Taking column No.

magic squares may be conas an example, Fig. 80 shows an

MAGIC SQUARES.

The above operation may be varied by reversing the horizontal columns of the key square Fig. 82 right and left as shown in Fig.
two
ancTthen forming the magic square given in Fig. 85. In this way different magic squares may be derived from each combination.

30 o

30

6
/z /z
/z
/s

/z

6
30
30

MAGIC SQUARES.
and eighth horizontal columns but cannot be employed for the second, third, sixth or seventh columns, and so
for the
first,

fourth, fifth,

forth.

Fig.

88 suggests half a dozen combinations which

/

will

form

42

MAGIC SQUARES.

adding these key numbers to the ^piiut^ numbers in Fig. 89 gives the popfc'ict magic square of 8 X 8 shown in Fig. 91. The latter will
be found identical with the square which

may

be written out directly

from diagrams
No.
i.

in Fig. 52.

No.

2.

No.

3.

No.

4.

No.

5.

No.

6.

MAGIC SQUARES.

43

63 61
/o

3'f

22
26~

39
2

3Z

33 3/ 60
24
2/

20

? 26 40 46 7
6-4

Totals

= 260.

60

64

Fig. 91.

44
Fig. 95

MAGIC SQUARES.
shows another

8X8 magic

square which

is

constructed

by combining the
square
in Fig.

primary square

in Fig.

89 with the

primary

changing the latter to a bey square in the manner before described? This magic ~square~~miay also be directly
constructed from diagram No. 4 in Fig. 54.
It
is

93 after

evident that an almost unlimited

number

of different

8X8

their application to the formation of other

magic squares may be made by the foregoing methods, and and larger squares is so
it

obvious that

will be unnecessary to present

any further examples.

COMPOUND MAGIC SQUARES.

these highly ingenious combinations
of.

Hermann

Schubert.

They may

be described as a series of small magic squares arin

ranged quadratically

magic square order.

is

The

9X9

square shown in Fig. 96

the smallest of this class

that can be constructed

and

it

consists of nine
i

3X3

sub-squares

arranged in the same order as the numerals

to 9 inclusive in the

3X3

square shown in Fig.

I.

The

first

sub-square occupies the

6?
4-f 6,7

SO

7*

26
2/
22

23
20

33

43

Totals

= 369.

3S
30 32 3V
3/

s-o

7* 7* '7

/o

77 7*
7*

/z
/3
/t

23 7*

Fig. 96.

middle section of the

contains the numbers

first
I

horizontal

row of sub-squares, and

I.

it

to 9 inclusive arranged in regular

magic

square order being a duplicate of Fig.

The second sub-square

MAGIC SQUARES.
is

45

located in the right

it

hand lower corner of the third horizontal row

contains the

of sub- squares and

numbers 10

to 18 inclusive ar-

ranged in magic square order, and so on to the last sub-square which occupies the middle section of the third horizontal row of

Sf if

23 -3V 22 32
44ft,

7*
7*
20 J/ '7

32

SS
SZ

JO

6s

Totals

= 369.

40
2t 33
7*-

7
7*

'

/S

3*

60 7'

S3 7*
Fig. 97-

/*&>

9*
9*
ft,

fv f7

"9
/23

fS 9* 9' SS
/2S /3
9-3

/2f S/S
33

6s 79 7*
7* 7
7'

97
Totals

42 43 3?
4S 77
63
60
62,

7*
fe /o?
/43

"7
/32 '7
3/

/*/

= 870.

30
23

S?

S? S3

/3f

3?

Z/
JZ,

Fig. 98.

numbers 73 to 8 1 inclusive. This peculiar arrangement of the numbers I to 81 inclusive forms a magic square in which the characteristics of the ordinary
sub-squares, and which contains the

46

MAGIC SQUARES.
square are multiplied to a remarkable extent, for whereas in

9X9

the latter square (Fig. 97) there are only twenty columns which

sum up to 369, in the compound square of 9 immense number of combination columns which
This
first
is

X9

there are an

yield this

amount.

evident from the fact that there are eight columns in the
1

sub-square which yield the number

also eight

columns

in

the middle sub-square which yield the

umns
15

in the last sub-square

which sum up

number 123 and eight colto the number 231 and

+
S3

123

+ 231 =369.

MAGIC SQUARES.

47

CONCENTRIC MAGIC SQUARES.

Beginning with a small central magic square it is possible to arrange one or more panels of numbers concentrically around it so that after the addition of each panel, the enlarged square will still
retain

magic

qualifications.

Either a

3X3
its

or

34X4

nucleus, and the square

will obviously

magic square may be used as a remain either odd or even,

according to

beginning, irrespective of the

it.

number of panels
center square will

which may be successively added to

The

/O

48

MAGIC SQUARES.

The
of 5

smallest concentric square that can be constructed

is

that

5,

an example of which
center square of 3
I,

is illustrated in

Fig. 99.

The

3 begins with 9 and continues, with

increments of

up

to 17, the center

with the general rule for a 5

number being 13 square made with

in

accordance

the series of

-Dtaaonal

Colu-rvnj

/X/ IRiiel

et.

JXJ

/O ft

Fig. 109.

Fig. 108.

'S<

6
/^>0

Fig

107.

Fig. 106.

MAGIC SQUARES.
other twelve numbers in the panels
relative positions of the nine
is

49
in Fig.

shown

101.

The
square

numbers

in the central

3X3

cannot be changed, but the entire square

may

be inverted or turned

one quarter, one

$

half, or three quarters

around, so as to vary the

Colu

X9

77

J7
2S

Fig. 113.

77
Fig.

in.
Fig. 112.

TOTALS

3X3 square 123, 5X5 square 205, 7X7 square 287, 9X9 square 369.

5O

MAGIC SQUARES.
square
is

3X3
right.

turned around one quarter of a revolution to the

Several variations

may

also be

made

in the location of the panel

numbers, an example being given

in Figs. 103, 104,

and

105.

Many

^uttttet-f in/

Column
/

6x6
o

&X

<2

tS^ftar-e.,

>

119.

Fig. 118.

2S
//

10

23

3/2,
Fig. 117.

J2

jr

<3<f

Fig. 120.

Totals of 4 Totals of 6

X4 X6

= 74. square = in.

square

other changes in the relative positions of the panel numbers are

selfevident.

One
is

of

many

variations of the 7

7 concentric magic square

shown

in Fig.

no.

The

3X3

central square in this

in order to

example

is

started with 21

and finished with 29

comply with the

MAGIC SQUARES.
general rule that 25 must occupy the center
that includes the series of
cell in

51

a 7

7 square
for the

numbers

to 49.

The numbers

two corner diagonal columns are indicated in their proper order by small circles in Fig. 106, and the arrangement of the panel numbers
is

given in Figs. 107, 108, and 109.

IN
J)taoonitc Columnsi/ristnferj ir

As

a final example of an

6x6 F i ane
'

^JJ-

JJ

X
'8
Fig. 123.
Fig. 122.

MAGIC SQUARES.
indefinitely

and used for the construction of larger odd magic

is

squares of this class.

The

smallest even concentric magic square

that of 6

X 6,

of

Fig. 125.

Fig. 129.

Totals of 4 Totals of 6
Totals of

X4 X6 8 X 8

= 130. square = 195. square = 260.

square

which Fig. 120

is

an example.

The development

of this square

may The

be traced in the diagrams given in Figs. 117, 118, and 119.

center square of 4

X4

is

perfect, but after the panel is

added

MAGIC SQUARES.
the enlarged square
122,
123,
\

53
Figs. 121,

as already noted.

and 124

illustrate

another example of this square with

diagrams of development.

&
QlOO

6*6
0*2

wo

10

9S
82.

Fig. 134.
26
4>Z

Fig. 13368

SS

Fig. 132.

/J

/o

6
Fig. 130.

Fig. 131.

SO 22 23 77 7* 26 9 *? 62
2S
6/ 7*

TOTALS

93 7* 6s 43 29 40
I

4X4 6X6 8X8

10

10

= 202 square = 303 square = 404 square = 505

square

7'

SO S2

/6
6* 39

7 fs 6s 7

7* 20
96

*
9Z
Fig. 135.

2* fz
/s

concentric square of 8

125, 126, 127,

132, 133, 134,

X 8 with diagrams are given in Figs. and 129, and one of 10 X 10 in Figs. 130, 131, 128, and 135. It will be seen that all these larger squares

54

MAGIC SQUARES.
in a

have been developed

very easy manner from successive expan-

sions of the diagrams used for the

6X6

square in Figs. 117, 118,

and

119.

The

rule's

governing the formation of concentric magic squares

difficult,

have been hitherto considered somewhat

diagrams, their construction in great variety

but by the aid of and of any size has

been reduced to an operation of extreme simplicity, involving only the necessary patience to construct the diagrams and copy the numbers.

GENERAL NOTES ON THE CONSTRUCTION OF MAGIC SQUARES.

There are two variables that govern the summation of magic squares formed of numbers that follow each other with equal increments throughout the
1.

series, viz.

The

Initial,

or starting number.
increasing number.

2.

The Increment, or
these

When
easily

two variables are known, the summations can be determined, or when either of these variables and the suminteresting problem in this connection

mation are known, the other variable can be readily derived.

The most
will therefore

is

the constructhis
is

tion of squares with predetermined summations, and

subject
familiar

be

first

considered, assuming that the reader

with the usual methods of building odd and even squares.

If a square of 3

is

constructed in the usual manner, that

is,

beginning with unity and proceeding with regular increments of I, the total of each column will be 15.

MAGIC SQUARES.

55

MAGIC SQUARES.

--

i) in the

above equation, and a table of these constants

all

is

therefore appended for

squares from

3X3toi2Xi2.
Const.

Squares

=K

3X3 4X4 5X5 6X6 7X7 8X8 9X9

10
11

12

30

60
105

168

252

360
495

12

X X X

10
ii

660 858

12

When

using the above constants the equation will be

EXAMPLES.

What
I

initial

number

is

required for the square of 3

3,

with

as the increment, to produce 1903 as the

summation?

Transposing the

last

equation

S
or
(i

0K =A,

12)

MAGIC SQUARES.

57

47? t

Totals

1903.

Fig. 140.

Also to a square of 5
T

5,

93

( i

X 60) = 368
J 6e. 6

= Initial

No.

3*7.6 AS?. 6

Totals

1903.

377.6

3S6.6

370.6

363.6 376.6
Fig.
141.

And

for a square of

6X6.
105)

1903- (IX

=
30/

= Initia

333

305-^
322 r 32/
3/3

Totals

JATf

= 1903-

302. f
Fig. 142.

The preceding examples

built

illustrate the construction of

i,

squares

up with progressive increments of

but the operation

may

be

varied by using increments that are greater or less than unity.

EXAMPLES.

What

initial
3,

number must be used

to

in a square of 3

3,

with

increments of

produce a summation of 1903?

MAGIC SQUARES.
Applying the equation given on page
instead of
i,

56,

but making

ft

we have:

622 1 / 3

is

therefore the

initial

number and by using

3,

this in a

3X3

square with progressive increments of

the desired results

are obtained.

Totals

1903.

Fig. 143.

To

find the initial

number with increments of

10.

Totals

1903.

Fig.

144.

Or

to find the initial

number with increments

12)

of

/8

'903

-(VsX

_ 6 ^ _ Inidal
63ST

NQ

633
Totals

1903,

Fig. 145-

These examples being sufficient to illustrate the rule, we will pass on another step and show how to build squares with predeter-

mined summations, using any desired

increment.

initial

numbers, with a proper

MAGIC SQUARES.

59

EXAMPLES.

What
wherein
I

increment number must be used in a square of

is

3X3,

the initial

number and 1903

the desired

summation?

Referring to equation

on page 56 and transposing, we have

A K
3)
12

ft

= increment,

or

-dX

= IS8 Y, = In crem ent.

,

Starting therefore with unity and building up the square with

successive increments of i58 1 / 3

we

obtain the desired result.

Totals

1903.

Fig.

146.

When
Thus
if

it

is

desired to start with any

number

larger or smaller

than unity, the numbers in the equation can be modified accordingly.

is

selected as an initial

number, the equation

:

will be

1903

(4
12

3)

57 Vie

Increment.

Totals

1903.

Fig.

147-

With an

initial

number of 1 / 3

1903

(Va X_3l

I5 gi^

ncrem ent.

792,

Totals

1903.

Fig.

148.

6o
It is

MAGIC SQUARES.
thus demonstrated that any
initial
is

number may be used

than one-third of the
less

providing (in a square of 3

summation.

In a square of 4

X 3) X4
case,

it

less

it

must be

than one-fourth

of the summation, and so on.

To
number

illustrate

an extreme
3 square and

we

will select

in a 3

find the increment

634 as an which will

initial

result

in a summation of 1903.

1903

-(634x3)
12

= lf = Increment
T

Totals

1903.

Fig. 149.

the formation of magic squares with summations by the use of proper initial numbers predetermined and increments, it only remains to show that the summation of any

Having now considered

square

may

be found,

when

the initial

number and

the increment
56, viz.
:

are given, by the application of the equation

shown on page

EXAMPLES.
Find the summation of a square of 3
number, and 7 as the increment.
(5

3 using 5 as the initial

3)

(7

12)

= 99 = Summation.
Totals

r*

= 99.

MAGIC SQUARES.

61

62

MAGIC SQUARES.

What
12
is

increment must be used

in

a square of 3

3 wherein

the initial

number and
12
12

12 the required

summation?

(12X3)

= Increment.

-/6

MAGIC SQUARES.
great variety of irregular sequences and
squares.
still

63

produce pe^fet magic

Referring to Fig. 40 as an example,

squares

many

different

5X5

may

be formed by varying the sequence of the five groups,

and also by changing the arrangement of the numbers in each group. Instead of writing the five diagonal columns in Fig. 40 with
the

numbers

to 25 in arithmetical order thus:

a.
b.
C.
i

6
II

7
12

8
13

9
14 19

10
15

d.
e.

16

17

18

20
25
d,

21

22

23

24

they
the

may

be arranged in the order b e c a square shown

in Fig. 17.

which

will develop

5X5

Other variations may be made by re-arranging the consecutive

numbers

in

each group, as for example thus:

a.
b.
c.

4
.9

6
ii

8
13

7
12

10
15

14
19

d.
e.

16
21

18

17

20
25

24

23

22

The foregoing may be considered

ways
of grouping
in

as only suggestive of

many

produced

numbers by which poi'foct magic squares may be great variety, which however will be generally found
numbers arranged
in similar

to follow the regular constructive rules heretofore given, providing

that these rules are applied to series of

consecutive order.

f en

MAGIC CUBES.
rTA
-*-

HE
may

curious and interesting characteristics of magic squares

be developed in figures of three dimensions constituting

magic Cubes of odd numbers may be constructed by tinuous process, and cubes of even numbers may be
aid of geometrical diagrams.

cubes.

direct
built

and con-

up by the

In each case the constructive meth-

ods resemble those which were previously explained in connection

with odd and even magic squares.

magic cubes, odd or even, aa^fhat all whether running from the top ofl^rcube to the bottom, frorrTtke front to the back, or from opi*^ide to the other,
characteristics of
,

amount, alsotj^rfthe four diagonal columns which unite the eigh^prners^^ffne cube and tho two corner
to

should

sum up

mName

the enbc should

sum up

to the

same amount as the stra^mcolumi^ Furthermore, in magic cubes the sun^^any two numbeN^hat are located
to each other

all

odd

in cells

and equicu^nt from the center diametrically ophite of the cuhidmould equal the sum of the first and lasN^umbers of the sen^^ised, and the center cell must contain the middlermmber of
ie

series.

OBD MAGTC CUBES. Of OPD N C/M_6J^5

The
smallest magic cube
tliis

is

naturally
cu.Dcil).lc

3X3X3.
of
lllan^*

seven numbers in

culjc iirc

different

MAGIC CUBES.

Fig. 157
III,

shows one of these cubes, and

in

columns

I,

II

and
it

Fig. 158, there are given the nine different squares which

contains.

In this cube there are twenty-seven straight columns,

two diagonal columns in each of the three middle squares, and four diagonal columns connecting the eight corners of the cube, making The in all thirty-seven columns each of which sums up to 42.
-4- i

center

number

is

also 14 or
is

and the sum of any pair of gee-

in metrically

opposite numbers

28 or n 3

i.

JL-

of Hit

Fig.

157-

In describing the direct method of building odd magic squares, many forms of regular advance moves were explained, including
right

and

left
It

diagonal sequence, and various so-called "knight's

also

moves."

was

shown

that the order of regular advance

was

periodically broken by other well defined spacings which were termed "breakmoves." In building odd magic squares, only one form of breakmove was employed in each square, but in the con-

odd magic cubes, two kinds are required in each cube which for distinction may be termed n and n 2 breakmoves respecIn magic cubes which commence with unity and proceed tively.
struction of

with increments of
tiple

i,

the n 2 breakmoves occur between each mul-

of n 2 and the next following number, which in a 3

X3X

cube brings them between 9 and

the
ficst

10, 18 and 19, and also between

and

last

numbers of the

series,

27 and

i.

The n breakmoves

66
are

MAGIC CUBES.

made between
and 24 and

all

other multiples of
7,

n,

which

in the

above case
16,

brings them between 3 and 4, 6 and

22, 25.

12

and

13, 15

and

21 and

With

this explanation the rules for building

i

the magic cube

shown

in Fig.

may now

be formulated, and for

is

convenience of observation and construction, the cube

divided

horizontally into three sections or layers, each section being

shown

separately in
It

Column

i,

Fig. 158.
is

may

be mentioned that when a move

the top square
it is

to be continued up-

ward from

carried around to the bottom square,

THREE SQUARES FROM TOP TO BOTTOM COLUMN I.

-

THREE SQUARES FROM FRONT TO BACK COLUMN II.

THREE SQUARES FROM LEFT TO RIGHT COLUMN III.

/o

MAGIC CUBES.
n breakmove.

67

One

//-

breakmove.

downward right-hand diagonal in next square down (from last entry). Same cell in next square down (from last
cell

in

entry).
If
it is

desired to build this cube from the three vertical squares

to

from front

back of Fig. 157, as shown

in

Column

II,

Fig. 158,
I

the directions will then be as follows: commencing with

in the

middle

cell

of the upper

row of numbers
cell

in the

middle square,

Advance move.
11

breakmove.

One One cell

up

in

next square up.

in

in

downward right-hand diagonal

next square up.

;r

breakmove.

tt^tr<
TABLE
I.

68 n 2 breakmove.
Five variations

MAGIC CUBES.

One

cell

down

in

same square (as

last entry).

may
I

be derived from this cube in the simple

way

illustrated in

Table

on the preceding page.

27 inclusive in Fig. 159, and change the
i

Assign three-figure values to the numbers

terms of
i,

to

2,

3 as given in Table

I,

numbers

in the three squares in

Column

I,

Fig. 158, to their cor-

responding three-figure values, thus producing the square shown in

Fig. 160.
It is

evident that

if

the arrangement of

numbers

in the

Column I were unknown, they could be readily from Fig. 160 by the translation of the three-figure values produced into regular numbers in accordance with Table I, but more than
three squares in

Mt'Jtllt,

MAGIC CUBES.

The combination
are

of

i, 2,

3 being given in

normal order

in the

original cube, the five cubes formed from the other combinations

shown

in Figs. 161-165.

These magic cubes may also be constructed by the direct method

in

accordance with the annexed directions.

MAGIC CUBES.

The

analysis of the

numbers

in Fig. 157
2, 3,

and Fig. 166

in Figs.

into their

three-figure values in terms of

167,

i,

as

shown

160 and

makes

clear the curious mathematical order of their arrangeis

ment which

not apparent on the face of the regular numbers as

FIG. 157

DIRECTIONS FOR CONSTRUCTING THE 3X3X3 MAGIC CUBE SHOWN IN AND FIVE VARIATIONS OF THE SAME.

COMBINATION

MAGIC CUBES.
column there
is

the

same

triple

repetition.

Furthermore,
to 42,
if

all

the

diagonal columns in the cubes which

their analyses in Figs. 160
lar

sum up

followed into

and 167

will also

be found to carry simi-

repetitions.

brief study of these figures will also disclose

other curious mathematical qualities pertaining to their intrinsic

symmetrical arrangement.

The next odd magic cube

shows one of
its

in order is 5

5,

and Fig. 168

it

many

possible variations.

For convenience,

is

divided into five horizontal sections or layers, forming five

5X5

squares from the top to the bottom of the cube.

Commencing with
A

in the first cell of the

middle horizontal

MAGIC CUBES.
50 straight columns summing up to 315 10 corner diagonal columns summing up to. 315 40 sub-diagonal columns summing up to.... 315 Total loo columns having the same summation.
.

DIRECTIONS FOR CONSTRUCTING THE 3X3X3 MAGIC CUBE SHOWN IN AND FIVE VARIATIONS OF THE SAME.

FIG. 166

COMBINATIONS

MAGIC CUBES.
315 50 straight columns summing up to 6 corner diagonal columns summing up to 315 20 sub-diagonal columns summing up to.... 315 Total 76 columns having the same summation.
. .

73

In the five vertical squares from right to

as in the last case, 76 columns which
all

left

of cube, there are,

sum up

to 315.

In the com-

plete cube there are also^four^diagonalj

rjjrht

immrr"Yrrnt-"'Tim up tn

~y

67

74
Select the
belozv
it

MAGIC CUBES.

fey-number which

is

nearest to the cell-number, but

in value.

Then

write
in

down
is

i.

The

section
>*<-**
T

number

which the key-number

__

found.

2.
3.

The primc-number over the -ke^c-number, The difference between the ke^t-number and
will thus

the cell-number.

Three figures

be determined which will represent the

required three-figure value of the cell-number.

Examples.
below
in value

The

first
is

number

in the first

row of

the upper

square in Fig. 168

it

67.

The

is

65

in section 3

nearest kgy-number to this and under the p*we number 4 and the
II.

TABLE

Sectio

/O

/s

20

4-0

SO

60

6s
90

7*

te

//O
Fig.
169.

difference between the fccy

number and the cell-number

is

is 2.

The

three-number value of 67

therefore

3. 4.

2.

Again, the fourth

it

number
in value

in the
is 5

same row
I

is 10.

The

nearest _hpy-number but below

j^iiin,

in section

under the

number

2,

and the

differ-

ence between the feey-number and the cell-number

'/ .Xligure value of 10
is

is 5.

The

three-

therefore
all

i. 2.

5.

By

these simple operations

the three-figure values of

the cell-numbers in the

5X5X5

may be quickly determined, and by the system of transposition previously explained, five variations of this cube may
cube in Fig. 168
be constructed.

MAGIC CUBES.

75

The

shorter

method of building these

5X5X5

cubes by the

direct process of filling the different cells in regular order with

consecutive numbers may, however, be considered by some to be

preferable to the

more roundabout way.

(See directions

in

the

following table.)

DIRECTIONS FOR CONSTRUCTING THE 5X5X5 MAGIC CUBE SHOWN IN AND FIVE VARIATIONS OF THE SAME.

FIG. 168

COMBINATIONS

76

MAGIC CUBES.
in the center cell of the

middle number of the series (63) appearing

cube according to rule. Odd magic cubes

may

be commenced in various

cells

other

than those shown in the preceding pages, and they

may

be built

up with an almost infinite number of variations. It would, however, be only superfluous and tiresome to amplify the subject further, as the examples already submitted cover aiLthe important points of
construction, and

may

readily be applied to further extensions.

<$/

/OS

zo 96
/9 /oo

so

"7
7*
7'
TOP SQUARE.
2.

30

J-J

62,

'7

//*

6/

7 Sf
BOTTOM SQUARE.

JJ

7*
S/3

77
s/z /a/

60

so J/

/oi
/J

T 7

26

JO

66 21

/os $6 /z 33
Fig. 170.

odd magic cubes larger than may be constructed by the directions which govern the formation of 3 X 3 X 3 and 5 X 5 X 5 cubes

Any

sizes of

5X5X5

MAGIC CUBES.Of
Magic cubes of even numbers may be
built

by the aid of geo-

metric diagrams, similar to those illustrated in the preceding chapter,

which describes the construction of even magic squares.

MAGIC CUBES.
Fig. 171

77

shows one of the many possible arrangements of a

is

4X4X4
viz.,

cube, the diagram of which

given in Fig. 172.

There are fifty-two columns in this cube which sum up to 130, sixteen vertical columns from the top of the cube to the
FIG. 170

DIRECTIONS FOR CONSTRUCTING THE 5X5X5 MAGIC CUBE SHOWN IN AND FIVEATARIATIONS OF THE SAME.

COMBINATIONS

MAGIC CUBES.
numbers, which are diametrically opposite to each other and equidistant from the center of the cube also equals 65 or n* i.

Another feature of

this

cube

is

that the

sum

of the four

num-

bers in each of the forty-eight sub-squares of 2

It

is

130.

has been shown in the chapter on "Magic Squares" that the

(Top.)
Section
I.

Section II

J2
Section III.

23

20 4-6

'7

Section IV.

sz

(Bottom.)

6/
Fig. 171.

Totals

130.

square of 4

X4

could be formed by writing the numbers

to 16

in arithmetical order, then leaving the

numbers
the other

in the

two corner
to their

diagonals unchanged, but changing

2 complements with 17 or n

all

numbers

-\- I.

It will

be noted in the magic cube

first

of

4X4X4,

given in Fig. 171, that in the

and

last

of the

MAGIC CUBES.
four sections (I and IV) this rule also holds good.

79
In the two
in the
all

middle sections (II and III) the rule

is

reversed

the

numbers
-(- I,

two corner diagonals being complements with 65 or n 3 the other numbers in arithmetical order.
Fig. 173

and

numbers

to

shows four squares or sections of a cube, with the 64 written in arithmetical order. Those numbers
Fig.
171

that occupy corresponding cells in

circles.

are enclosed within

numbers in Fig. 173 are changed to their complements with 65, the total arrangement of numbers will then be the same as in Fig. 171. In his interesting and instructive chapter entitled "Reflections
If all the other

on Magic Squares"* Dr. Paul Carus gives a novel and ingenious analysis of even squares in different "orders" of numbering, these orders being termed respectively o, ro, i and ri. It is shown that
the

two magic squares of 4

X4

(in the chapter referred to) con-

'7

8o
174
is

MAGIC CUBES.
a 4

X 4X 4
line,

cube which

is

constructed by writing the four

numbers

that are contained in the

2X2

sub-squares (Fig. 171) in

a straight

and there are many other possible transpositions

relative order of the

which

will

change the

numbers, without de-

stroying the magic characteristics of the cube.

Section

I.

6Z

(Top.}

/.s

Section

II.

30
Jf 2S J/ J3

JZ
Section III.

;*

MAGIC CUBES.

81
in order.

The
shows

8X8X8
first

this

Fig. 176 cube divided, for convenience, into eight horizontal layers

magic cube follows next

or sections, and Fig. 177 gives the diagrammatic order of the

bers in the
built

num-

and eighth

sections, the intermediate sections being

in arithmetical order.

from similar diagrams, numbered

82

MAGIC CUBES.

An
cube
sections

examination of Fig. 176 will show that,

it is

like the

4X4X4
In

in Fig. 171
I,

built

up

of o and ro numbers exclusively.

cells in the

IV, V, and VIII, the

corner diagonal columns,

in definite geometrical
all

and

in certain other cells

which are placed

numbers, while

relations thereto, contain o

the other cells con-

as/

zs-o 2.6*.

/2V
//f

tzt

166
2^0 27*
2f/
2.42

396 3V?
/o6
S00

233 226

97

Z9J Z/J 2/g 296

#6

33

2/3
JO 6

302 306 209

204.

ft

fi

20S

7* 7*
63
6f

7*

J/7 /3S '3# 32O

Section V.

Section VII.
\

322 323

326 327 /Sf

3Z2 J33
'7 f
JT6

6*
#62,

J23
'7

463

7'

/ o

/6f

4/3
S
~7

3/ 2?
37 6

30

26

36/

36*

/6

/z S03

S-oZ

J3&

/33

JfS.

47Z

Section VI.

Section VIII.

(Second Part.)
Fig.
176.

tain ro

numbers.

In sections

II,

III,

VI, and VII, the relative

positions of the o and ro numbers are reversed.

By
numbers

noting the symmetrical disposition of these two orders of

in the

different sections, the

cube

may

be readily conFig.

structed without the aid of any geometrical diagrams.

178

shows sections

and

II of Fig.

176

filled

with o and ro symbols

MAGIC CUBES.
without regard to numerical values, and the relative symmetrical

arrangement of the two orders

is

therein plainly illustrated.

This

Fig.

177.

"o" numbers.

~V'ro" numbers.

O O o
o
clear

oo oo
oo oo
and

oo

oo oo 00

oo oo
00 oo

o o o o
Fig. 178.

o o o o

oo

oo oo

oo

00 oo
oo oo

oo
00 oo

o o o o

00

lucid analysis, for

which we are indebted to Dr. Carus,

reduces the formation of a rather complicated numerical structure

to an operation of the utmost simplicity.

84

MAGIC CUBES.

In this cube there are 192 straight columns, and 4 diagonals columns (which unite the eight corners of the cube) each of which sums up to 2052 also 384 half columns and the same number of
;

2X2

sub-squares each of which has the summation of 1026.

It

will also be seen that the

sum

of any

two numbers, whieh are

lo-

cated in cells diametrically opposite to each other and equidistant

x from the center of the cube, is 513 or n* As the writer lias been unable to construct the
-

6X6X0 magic
X
10

cube no attempt

will-

be made in this book to produce-a-LCLX 10

or any larger cube of this peculiar class.

The

12

12

12 cube

and

all

larger ones that are formed with multiples of 4 will naturally

resemble the

8X8X8

cube and will be equally easy to construct.

GENERAL NOTES ON MAGIC CUBES.

Magic cubes may be constructed having any desired summations by using suitable initial numbers with given increments, or
by applying proper increment^to given
* *
*
initial

numbers.

The formula
is

for determining the summations of

in connection

magic cubes

similar to that

which was given

with magic squares

and may be expressed as follows:

Let:

A = initial
fi

number,

then

if

= increment, n = number of cells in each column of cube, S = summation A = and = i:

;

J3

If

and

f$

are

more or

less

than unity, the following general

formula

may

be employed:

An + p^
To

(n

i)

=S.
i)

shorten the above equation,

- (n 5

may

be expressed

as a constant

(K)

for each size of cube as follows:

MAGIC CUBES.

Cubes.

Const.

= K.

3X3X3 4X4X4
5X5X5--:

39
126

3io
!.

6X6X6 7X7X7 8X8X8 9X9X9

10

645

H97
2044
3 2 76

10

10

4995

When
(i)

using the above constants the equation will be

Aw + /?K
S

= S,

An

(2)

(3)

EXAMPLES.

What
with an

increment number

is

required for the cube of 3

X3X

initial

number

of 10 to produce summations of 108?

in figure values:

Expressing equation (2)

108

(IQX3)

Jo 20
/o
1

4-2

a*

86

MAGIC CUBES.

SO

MAGIC CUBES.

ment of numbers and

in various

minor

features, while retaining the

common

characteristic of having similar

is

column

values.

An

ex-

ample of this differentiation

seen in the interesting "Jaina" square

w/.z

88

MAGIC CUBES.
Another
class
is

illustrated

in

the "Franklin

Squares," and

these can also be expanded into cubes constructed on the same

general principles.

The

subject of magic squares and cubes

indefinitely extended.

is

indeed inexhaustible

and may be

The

philosophical significance

of these studies has been so ably set forth by Dr. Carus that the
writer considers
it

unnecessary to add anything in this connection,

but he trusts that the present endeavor to popularize these interesting problems

may some

time lead to useful results.

CHAPTER

TIL

THE FRANKLIN SQUARES.

following letter with magic squares of 8
is

X8

and 16

16

copied from "Letters and papers on Philosophical subjects

S.,"

by Benjamin Franklin, LL. D., F. R.

in

a work which was printed

London, England,

in 1769.

FROM BENJAMIN FRANKLIN

ESQ. OF PHILADELPHIA,

To
DEAR
SIR:

PETER COLLINSON ESQ. AT LONDON.

According to your request I now send you the arithmetical curiosity of which this is the history.

Being one day

in the country at the

house of our

common
French

friend, the late learned

Mr. Logan, he showed me a

if I

folio

book

filled

with magic squares, wrote,

forget not by one Mr.

which he said the author had discovered great ingenuity and dexterity in the management of numbers; and though several
Frenicle, in

other foreigners had distinguished themselves in the same way, he

did not recollect that any one Englishman had done anything of the

kind remarkable.
I said it

was perhaps a mark of the good sense of our mathe-

maticians that they would not spend their time in things that were

He merely difficiles nugce, incapable of any useful application. answered that many of the arithmetical or mathematical questions
publicly

proposed in England were equally trifling and useless. Perhaps the considering and answering such questions, I replied, may not be altogether useless if it produces by practice an habitual

THE FRANKLIN SQUARES.

readiness and exactness in mathematical disquisitions, which readi-

ness may, on
says he,
fessed to

many
the
that in

occasions be of real use.

In the same
I

way

may
him
I still

making

of these squares be of use.

then conleisure
I

my

(which

think
in

younger days, having once some might have employed more usefully)
squares,
fill

had

amused myself

making these kind of magic

at
size
it,

and, at

length had acquired such a knack

that I could

the cells of

any magic square of reasonable

fast as I could write

with a series of numbers as such a manner that the sums

or
diagonal,
I

them,

disposed in

of

every
;

row,

horizontal,

perpendicular

should be

equal

but not being satisfied with these, which

looked on as com-

20

S'J

7'

Fig.

183.

mon and

easy things,
in

had imposed on myself more

difficult tasks,

and succeeded
properties,
in the

making other magic squares with a variety of

and much more curious.

He

then showed

me
;

several

same book of an uncommon and more curious kind

but as

I thought none of them equal to some I remembered to have made, he desired me to let him see them; and accordingly the next time

I visited

him,

carried

him a square of 8 which

I will

found among

my
its

old papers,

and which

now

give you with an account of

properties (see Fig. 183).

1.

The

properties are:

That every straight row (horizontal or vertical) of 8 numbers added together, makes 260, and half of each row, half of 260.
2.

That the bent row of 8 numbers ascending and descending

THE FRANKLIN SQUARES.

diagonally, viz.,

91

from 16 ascending
its

17 and every one of

etc.

parallel

and from 23 descending to bent rows of 8 numbers make 260, etc.,

to 10

And

lastly the four corner

numbers with the four middle numbers

24 2 '6
2S3 ZAO

4V

S3

/SJ

66

1*0

#7 /$$
676

ot

2,3

/*>

Z&4 2/4

&/Z 266
^J 2/2

Wt /$S
206

2*9

ZS S3

fi
2*6 ZfS

&*

3*? 228 />3

223

Fig. 184.

/2
//s
',1

to

X
XiA

Fig. 185.

make
at

260.

So

this

magical square seems perfect in

its

kind, but

these are not

all its

properties, there are 5 other curious ones which

'

some time

will explain to you.

Mr. Logan then showed me an old arithmetical book

in quarto,

92
wrote,
I

THE FRANKLIN SQUARES.

think by one Stifelius, which contained a square of
16

which he said he should imagine to be a work of great labour; but if I forget not, it had only the common properties of making the

same sum, viz., 2056 in every row, horizontal, vertical and diagonal. Not willing to be outdone by Mr. Stifelius, even in the size of my went home, and made that evening the following magical square of 16 (see Fig. 184) which besides having all the properties of the foregoing square of 8, i. e., it would make 2056 in all the
square, I

same rows and diagonals, had

just 16 of the
little

this added, that

a four-square hole

being cut in a piece of paper of such a size as to take in and

show

through

it

squares,

when

laid

on the greater

Fig.

186.

square, the

sum

of the 16 numbers so appearing through the hole,

likewise
after

wherever
'2056.

it

was placed on the greater square should

I sent to

make
some

This
it

our friend the next morning,

letter

who

days sent

back in a

with these words:

"I return to thee thy astonishing

"or most stupendous piece "of the magical square in which" ....

but the compliment

as well as
for I

is

too extravagant and therefore, for his sake,

Nor is it necessary, I ought not to repeat it. no question but you will readily allow the square of 16 make

my own

THE FRANKLIN SQUARES.

to be the

93

most magically magical of any magic square ever made

I

by any magician.

am

etc.

B. F.
in Figs. 183

It willjbe

seen that the squares

shown

and 184 are

not perfect according to the rules

^@

magic squares previously

Fig.

187.

given, but the interesting feature of their bent diagonal columns

calls for

more than passing

notice.

In order to
is

facilitate the

study

of their construction, a 4

X4

square

given in Fig. 185 which

presents similar characteristics.

94

THE FRANKLIN SQUARES.

The
dotted lines in this square indicate four bent diagonal col;

total of 34 three of these columns being and one being broken. Four bent diagonal columns may be formed from each of the four sides of the square, but only twelve of these sixteen columns have the proper totals.

umns, each of which has a

intact within the square

Adding

to these the eight straight columns,

we

find that this square

contains twenty columns with summations of 34.

The

4X4 "Jaina"

square contains sixteen columns which ordinary

sum up

to 34 while the

The

4X4 magic square contains only 8X8 Franklin square (Fig. 183)

contains

forty-eight

columns which sum up

to 260, viz., eight horizontal, eight perpen-

dicular, sixteen bent horizontal diagonals,

and sixteen bent perpenS9

60
Section
i.

/Z Z/

7 S6
Section
2.

43 42

(Top.)

2$

6z

THE FRANKLIN SQUARES.

sub-square
little is

95

514,

and there are also other curious features which a

study will disclose.

J '*

.SecK**

Fig. 189.

The Franklin Squares possess a unique and

in the

peculiar

symmetry
in their

arrangement of their

numbers which
is

is

not clearly observable

on

their faces, but

which

brought out very strikingly

90

THE FRANKLIN SQUARES.

geometrical diagrams as given in Figs. 186 and 187, which illustrate respectively the diagrams of the 4 4 and squares.

8X8

be readily constructed by expanding these diagrams and writing in the appropriate numbers. The cube of 4 4 4 an d its diagram are given as examples in Figs. 188 and 189, and it will be observed that the curious char-

Magic cubes

may

X X

acteristics of the square are carried into the cube.

AN ANALYSIS OF THE FRANKLIN SQUARES.

In The Life and Times of Benjamin Franklin, by James Parton, (Vol. I, pp. 255-257), there is an account of two magic squares, one

8X8,

the other 16

16,

which are given here

in Figs. 191

and

192.

= 260

II
130 130

260

= 260

130

260

Hh

= 260

=260 PROPERTIES OF FRANKLIN^S

Fig.
IQO.

= 260

=260

8X8

SQUARE.

Mr. Parton explains the

8X8

square as follows:
its

"This square, as explained by

"properties: every straight

contriver, contains astonishing

row

(horizontal or vertical) added to-

THE FRANKLIN SQUARES.

52

97

THE FRANKLIN SQUARES.

= 1028

2056

=F

= 2056

= 2056

= 2056

= 2056
L -H

2056

=1028
PROPERTIES OF FRANKLIN^S
Fig.
IQ3-

=2056

128

l6Xl6 SQUARE.

THE FRANKLIN SQUARES.

"gather makes 260, ana each half row half 260.

99

The bent row

viz.,

of

"eight numbers ascending and descending diagonally,

from 16

"ascending to 10,
"its parallel

and from 23 descending to 17, and every one of bent rows of eight numbers, makes 260. Also, the bent

2056

2056-4-128

~-jo-.fi

j n

= 2056
PROPERTIES OF FRANKLIN^S
Fig- 193 (con.)-

r= 20J-6

l6Xl6 SQUARE.

"row from 52 descending

"every one of
"Also, the bent
its

to 54, and from 43 ascending to 45, and bent rows of eight numbers, makes 260. parallel

row from 45

to 43, descending to the

left,

and from

IOO

THE FRANKLIN SQUARES.

17,

"23 to

descending to the right, and every one of

its

parallel bent

"rows of eight numbers, makes 260. Also, the bent row from 52 "to 54, descending to the right, and from 10 to 16, descending to
"the
left,

and every one of

its

parallel bent

rows of eight numbers,

"makes 260.

Also, the parallel bent rows next to the above-men-

tioned, which are shortened to three numbers ascending and three

from 53 to 4 ascending and from 29 to 44 "descending, make, with the two corner numbers, 260. Also, the two "numbers, 14, 61, ascending, and 36, 19, descending, with the lower
"descending,
etc.,

as

"four numbers situated like them,

"ascending, makes 260.

viz., 50, i,

descending, and 32, 47,

And,

lastly,

the four corner numbers, with

"the four middle numbers,

make
all

260.

"But even these are not

"square.
Its contriver

the properties of this marvelous

it

declared that
;

has

'five

other curious ones,'

"discover

"which he does not explain if he can."

but which the ingenious reader

may

These remarkable

characteristics

which Mr. Parton enumerates

are illustrated graphically in the accompanying diagrams in which the

relative position of the cells containing the

numbers which make up

the

number

260,

is

indicated by the relation of the small hollow

squares (Fig. 190).

Franklin's

16X16
this

square

is

constructed upon the same principle

as the smaller, and Mr. Parton continues:

"Nor was
"squares.

the most wonderful

of

Franklin's magical

He made

one of sixteen

cells in

each row, which besides

"possessing the properties of the squares given above (the amount,

"however added, being always 2056), had

"able peculiarity
:

also this

most remark-

a square hole being cut in a piece of paper of such

it

"a size as to take in and show through

"squares,

just sixteen of the

little

when

laid

"bers, so appearing through the hole, wherever

on the greater square, the sum of sixteen numit was placed on the

"greater square, should likewise

make 2056."
Mr. Parton notes of the i6X

The
1

additional peculiarity which

is

6 square

is

true of the smaller square, that the

no more remarkable than the corresponding fact which sum of the numbers in any

A
B

C
D
E

G
H

IO2

THE FRANKLIN SQUARES.

combination of
its

2X2

cells

yields

130.

The

properties of the

larger square are also graphically represented here (Fig. 193).

A
follows

clue to the construction of these squares

:

may

be found as

We
letters

write

down

the

numbers

in numerical order

and

call

the

cells after the

precedent of the chess-board, with two sets of symbols,

and numbers.

We

call this

"the plan of construction" (Fig.

194).

Before

we

construct the general scheme of Franklin's square

we

will build

ciple,

which

We

will

up another magic square, a little less complex in prinwork for more complicated squares. simply intermix the ordinary series of numbers according
will be preparatory

to a definite rule alternately reversing the letters so that the

odd

rows are

in alphabetical order

and the even ones reversed.

In order

to distribute the

numbers

in

a regular fashion so that no combinastart with

i

tion of letter

and number would occur twice, we

in the

upper left-hand corner and pass consecutively downwards, alternating between the first and second cells in the successive rows,
thence ascending by the same method of simple alternation from
in the
i

lower left-hand corner.

We
in

have

now

the key to a scheme

It is

for the distribution of

first

numbers

an

8X8

magic square.

the

step in the construction of the Franklin

it

8X8

magic square, and

we

call

"the key to the scheme of simple alternation" (Fig. 195).

if

It

goes without saying that the effect would be the same

we

begin in the same

way

in the right-hand corners,

only

we must

;beware of a distribution that would occasion repetitions.

To

complete the scheme

we have

to repeat the letters, alternately

mvertiiig, their order

row

after row,

and the

first

two given

figures

must be repeated throughout every row, as they are started. The top and bottom rows will read i, 8; i, 8; i, 8; i, 8. The second

row from
7, 2.
3,

the top and also from the bottom will be

third

7,

7,
3,

2
6;

7, 3,

The
3,

row from

the top and bottom will be

5,

6;
In
the

6;

6; and the two center rows

4;

5,

4;

5,

4:

5,

4.
is

every line the

sum

of two consecutive figures yields

9.

This

second step, yielding the completed scheme of simple alternation

(Fig. 196).

THE FRANKLIN SQUARES.

The square
is

IO3

now produced by

substituting for the letter and

figure combinations, the corresponding figures according to the con-

secutive arrangement in the plan of construction (Fig. 197).

Trying the

results

we

find that all horizontal

260, while the vertical rows are alternately 260

rows sum up to 4. 4, and 260

The diagonal from the upper right to the lower left corner yields a sum of 260+32, while the other diagonal from the left upper The corner descending to the right lower corner makes 260 32.
the lower halves, and the

upper halves of the two diagonals yield 260, and also the sum of sum total of both diagonals is accordingly
520 or 2X260. 260

The sum

of the two left-hand half diagonals re-

sults in

right-hand side

16, and the sum of the two half diagonals to the makes 260+16. The sum of the four central cells

plus the four extreme corner cells yields also 260.

Considering the fact that the figures

to 8 of our

scheme run

up and down in alternate succession, we naturally have an arrangement of figures in which sets of two belong together. This binate peculiarity is evidenced in the result just stated, that the rows yield

sums which are the same with an

of an equal amount.

alternate addition

and subtraction

So we have a symmetry which is astonishing and might be deemed magical, if it were not a matter of intrinsic
necessity.

We

represent these peculiarities in the adjoined diagrams (Fig.

all

198) which, however, by no means exhaust

the possibilities.
re-

We

must bear

in

mind

that these
is

magic squares are to be

one another as

garded as continuous; that

to say, they are as if their opposite

if

sides in either direction passed over into

they

were joined both ways

in the

shape of a cylinder.

In other words

when we cross the boundary of the square on the right hand, the first row of cells outside to the right has to be regarded as identical with the first row of cells on the left; and in the same way the
uppermost or first horizontal row of cells corresponds to the first row of cells below the bottom row. This remarkable property of the square will bring out some additional peculiarities which mathematicians
cially

may

easily derive according to general principles

espe-

what was

stated of the

sum

of the lower and upper half-

104

THE FRANKLIN SQUARES.

U
260

+ 16

D =260

16

260

32

260 -f 32

260

2604

260+4

= 260
Fig. 198.

= 2 X 260
8X8

s=s

20

PROPERTIES OF

SQUARE BY SIMPLE ALTERNATION.

AB

A = 2056 8 B = 2066 + 8

= 2X2056

= 2056

= 2056
Fig. 199.

PROPERTIES OF

l6Xl6 SQUARE BY SIMPLE ALTERNATION.

THE FRANKLIN SQUARES.

diagonal of any bent series of

upward
opposite

or

running staircase fashion either downward to the center, and hence proceeding in the
cells

way to the other side. The magic square constructed according

to the

method of sim-

2056

128

= alternately 2056

64 and 2056

+ 64

LJ

= 2056 +
PROPERTIES OF

28

= alternately 2056
Fig.

U
64 and 2056

+ 64

l6X l6 SQUARE BY SIMPLE ALTERNATION.

IQ9 (con.).
is

pie alternation of figures

Franklin, but

we

however, the square of Benjamin can easily transform the former into the latter
not,

by slight modifications.

We

notice that in certain features the

sum

total of the bent

io6

THE FRANKLIN SQUARES.

THE FRANKLIN SQUARES.

4
5
6
7

10

11

12

13

14

15

16

io8
A1

THE FRANKLIN SQUARES.

THE FRANKLIN SQUARES.

A
1

no
1

THE FRANKLIN SQUARES.

THE FRANKLIN SQUARES.

Ill

diagonals represents regularities which counterbalance one another

on the right- and the left-hand side. In order we have to shift the figures of our scheme.

to offset these results

We

take the diagram which forms the key to the scheme of our

distribution by simple alternation (Fig. 195),

middle, turn the lower half upside down, giving the

and cutting it in the first two rows as

seen in Fig. 200 in which the heavy lines indicate the cutting. Cutting

then the upper half in two

(i. e.,

in binate sections),

and transposing

the second quarter to the bottom,

rangement of figures

in

we have the key to the entire arwhich the alternation starts as in the

scheme for simple alternation but skips the four center rows passing from 2 in the second cell of the second row to 3 in the first cell of
the seventh,

5 in the first cell,

and from 4 in the second and thence upwards

of the

cell

of the eighth passing to

in similar alternation, again

passing over the four central rows to the second and ending with 8
in the

second
in the

cell

first

row.

Then

the

same alternation

is

pro-

duced

four center rows.

It is

obvious that this can not start

first

in the first cell as that

i

would duplicate the

row, so

we

start

with

in the

second

cell

passing

down

uninterruptedly to 4 and ascending

as before

from

5 to 8.

A
means

closer examination will

in sets of two. sets of

show

that the

rows are

binate,

which

The

four inner numbers,

each,
i,

3, 4, 5,

6 and the

two outer

two numbers

2 and

7, 8,

are brought to-

gether thus imparting to the whole square a binate character (Fig.

202).

We
the

are

now

provided with a key to build up a magic square

after the pattern of Franklin.

We

have simply to complete

it

in

same way
line.

as our last square repeating the letters with their

order alternately reversed as before, and repeating the figures in

each

When we

insert their figure values

we have
in

a square which

is

not the same as Franklin's, but possesses

qualities (Fig. 203).

principle the same

To make our 8X8 Franklin square we must say, we preserve exactly

square of binate transposition into the

first

take

its

obverse square

that

is

to

the same order but holding the paper

112

..

THE FRANKLIN SQUARES.

with the figures toward the light we read them off from the obverse side, and then take the mirror picture of the result, holding the

mirror on either horizontal

side.

So

far

we have

still

our square

with the peculiarities of our scheme, but which lacks one of the
incidental
that he
characteristics

of Franklin's

square.

We

must notice

makes four
to 130

cells in

both horizontal and vertical directions

is

sum up
sets of

which property
each row. If

necessarily limited only to

two

four

cells in

we

write

down
sum

the

sum

of
set

3 _|_ 4 _|_ 5 _(_6+7+8=2Xi8, we

will find that the

middle

1+2+ 3+4+
,

5+6

is

equal to the rest consisting of the

of two extremes,

1+4, and 7+8. rows represented by the

In this
cordingly

way we

cut out in our scheme (Fig. 202) the

letters C,

D, E,

in either order

and

ac-

we

can shift either of the two

side.

first

or two last vertical

rows to the other

his square with

Franklin did the former, thus beginning

G4

in the left

upper corner as in Fig. 204.

We have

indicated this division by heavier lines in both schemes.

The
the

greater square of Franklin, which

same

is 16X16, is made after and the adjoined diagrams (Figs. 205-212) will fashion,

sufficiently explain its construction.

We
sess

do not know the method employed by Franklin; we posonly the result, but it is not probable that he derived his square

according to the scheme employed here. Our 16X16 square is not exactly the same as the square of
Franklin, but
it

belongs to the same

class.

Our method

gives the

key to the construction, and it is understood that the system here represented will allow us to construct many more squares by simply

pushing the square beyond its limits into the opposite row which by this move has to be transferred.

There

is

the same relation between Franklin's

16X16

square

and our square constructed by alternation with quaternate transposition, that exists between the corresponding 8X8 squares.
p. c.

C H A

PTE F?

REFLECTIONS ON MAGIC SQUARES.

A MATHEMATICS, *^* ophy, has always
article

especially in the field

where

it

touches philosS.

been

my

foible,

and so Mr.

W.

Andrews's

on "Magic Squares" tempted me to seek a graphic key to the interrelation among their figures which should reveal at a glance

the mvsterv of their construction.

THE ORDER OF FIGURES.

In odd magic squares, 3

3>

5>

7>

etc.,

there

is

no

difficulty whatever, as Mr. Andrews's diagrams

show

at a glance

(Fig. 213).

The

consecutive figures run up slantingly in the form

114

REFLECTIONS ON MAGIC SQUARES.

This cannot be done at once with both
its

shape of a cylinder.
opposite vertical
cess
is

its

two

and

two opposite horizontal

sides,

but the pro-

easily represented in the plane

all its sides,

extended on

must

treat

magic square and -on passing its limits on one side we the extension as if we had entered into the magic square
left
it.

by having the

on the side opposite to where we

over in a

If

we now

transfer the

figures to their respective places in the inside square, they are shoved

way which by

a regular transposition will counteract their

regular increase of counting and so equalize the

sums of

entire rows.

case is somewhat more complicated with even magic and a suggestion which I propose to offer here, pertains squares, to their formation. Mr. Andrews begins their discussion by stating
that "in pcrfuct magicr JCltuircs of thic clacc
it

The

is

necessary thEt the

ch column shall be the same amount, and also that the

y two numbers that are geometrically equidistant from the

center --'Of the square shall equal the
bers

sum

of the

first

and

last

num-

oLihe

series."

The
in a

smallest magic square of even

if

numbers

is,

of course,

4X4:

and he points out that 4

we

write the figures in their regular order

X4

square, those standing on the diagonal lines can remain

in their places,

while the rest are to be reversed so as to replace

its

every figure by

complementary to 17

(i. e.,

2 by 15, 3 by 14, 5 by

12, 9 by 8) the number 17 being the sum of the highest and lowest numbers of the magic square (i. e., n 2 -(- i). It is by this reversal

of figures that the inequalities of the natural order are equalized

again, so as to

make

the

sum

of each

row equal

to 34,

which

is

one

fourth of the

sum

total of all figures, the general

formula being

We

will

now

try to find out

more about the

relation

which the

magic square arrangement bears to the normal sequence of figures. For each corner there are two ways, one horizontal and one
normal sequence accordingly there are altogether eight possible arrangements, from which we select one as fundamental, and regard all others as mere
vertical, in in the

which figures can be written

variations,

produced by inverting and reversing the order.

REFLECTIONS ON MAGIC SQUARES.

115

As

the fundamental arrangement

left to

we choose

the ordinary

way

of writing from the

the right, proceeding in parallel lines

o.

downward.

We

call this

"the ordinary order" or

Its

reverse

proceeds from the lower right-hand corner toward the left, and line by line upward, thus beginning the series where the ordinary arrangement ends, and ending where it started. We call this order
"the reversed ordinary," or simply ro.

Another order

is

produced by following the Hebrew and Arabic

mode

of writing:

ing to the left,

we begin in the upper right-hand corner, proceedand then continue in the same way line by line
we

downward.
call briefly
i.

This, the inverse direction to the ordinary way,

The
on we

reverse order of

i,

starting in the lower left corner, pro-

ceeding to the right,

shall

and

line

by

line

upward we

call

n.

Further
fol-

have occasion to represent these four orders by the

:

lowing symbols

by

ro by

by

by

-|-.

n6
/

REFLECTIONS ON MAGIC SQUARES.

REFLECTIONS ON MAGIC SQUARES.

It will

117

be noticed that

i is

the vertical mirror picture of o and

is

ro of

ri,

and vice

versa. Further if the mirror

ri is

placed upon one

i

of the horizontal lines,

the mirror picture of o as well as ro of

and vice versa. There are four more arrangements. There is the Chinese way of writing downward in vertical columns as well as its inversion,
and the reversed order of both.
of
the

This method originated by the use

bamboo

strips as writing material in China,

two vowel sounds of the word "bamboo"

and we may utilize (viz., a and u) to name and u the

right,

the left and the right

downward

order, a the left

the reverse of the right ru and of the left ra, but for our present

purpose there will be no occasion to use them.

Now we

must bear

in

mind

that

magic squares originate from

the ordinary and normal consecutive arrangement by such transpositions as will counteract the regular increase of value in the nor-

mally progressive series of figures

and these transpositions depend

cells.

upon

the location of the several

All transpositions in the

cells of

even magic squares are brought about by the substitution

i,

of figures of the ro,

and

ri

order for the original figures of the

order, and the symmetry which dominates these becomes apparent in the diagrams, which present at a glance changes the order to which each cell in a magic square belongs.

ordinary or o

Numbers of

the

same order are grouped not unlike the Chladni

it

acoustic figures, and

seems to

me

that the origin of the regular-

ity of both the magic figures and this

phenomenon of
or
in

acoustics, is

due to an analogous law of symmetry. The dominance of one order o, ro,

even magic square,
selection
is

i,

ri,

each

cell

of an

simply due to a definite method of their

different orders of counting.

from the four

cell

Never can

a figure appear in a

where
i,

it

does not belong by right of some

regular order, either o, ro,

or n.
consists only of o

The magic square

of 4

X 4,

and ro

figures,

and the same rule applies to the simplest construction of even squares of multiples of four, such as 8 X 8, and 12 X 12.

There are several ways of constructing a magic square of 6 X 6. Our first sample consists of 12 o, 12 ro, 6 ri, and 6 i figures. The

REFLECTIONS ON MAGIC SQUARES.

12 o hold the diagonal lines.

The

12 ro

go

parallel

with one of

these diagonals, and stand in such positions that

square were diagonally turned upon

the 6
i,

itself,
i

and 6

ri figures.

And

again the 6

if the whole magic would exactly cover they and 6 ri also hold toward

each other places in the same

if

way corresponding

to one another;

the magic square were turned

onal, each ri figure

upon itself around the other diagwould cover one of the i order.

REFLECTIONS ON MAGIC SQUARES.

119

Fig. 217.

CHLADNI FIGURES.*
is

* The letter a indicates where the surface marks the place where the bow strikes the

touched with a finger

glass plate.

while In the four upper

;

120
verse of o which

REFLECTIONS ON MAGIC SQUARES.

is

ro represents one-half turn,

and

ri

the

first

and

third quarter in the whole circuit,

and

it

is

natural, therefore, that

in the

a symmetry-producing wave should produce a similar effect

magic
plate.

square to that of a note

upon the sand of a Chladni

glass

MAGIC SQUARES IN SYMBOLS.

The diagrams which
are offered here in Fig. 218 are the best

evidence of their resemblance to the Chladni figures, both exhibiting in their formation, the effect of the law of symmetry. The most

>'
8X8.
32 o and 32 ro.
10

icx

72 o and 72 ro.
ro.

SQUARES OF MULTIPLES OF FOUR.

Constructed only of o and

++
it

* + +

++ ++ ++'
+'
all

++
+
i,

8X8 SQUARES.
Constructed from
the orders, o, ro,

and

ri.

Fig. 218.

diagrams the plate has been fastened in the center, while in the lower ones has been held tight in an excentric position, indicated by the white dot

REFLECTIONS ON MAGIC SQUARES.

elegant

121

way

of rendering the different orders,

i,

ri, o,

and

ro, visible

at a glance,

would be by printing the

cells in

four different colors,

ANOTHER 8X8 SQUARE.

be noted that in this square the arrangement of the o symbols corresponds very closely to the distribution of the sand in the second of the Chladni diagrams. The same may be said of the two following figures, and it is especially true of the first one of the squares just preceding.
It will

8X8

*++
12 o, 12 ro, 6
i,

ri.

+*+
+
!'

+*+ ++* +*+

+
o,

40

40 ro, 10

i,

10

ri.

The reader
between

will notice that there is a remarkable resemblance the symmetry displayed in this figure and in the fourth

of the Chladni diagrams.

Fig. 218. (con.).

EXAMPLES OF 6X6 AND ioX 10 MAGIC SQUARES.

it

but for proving our case,

will

be

sufficient to

have the four orders

represented by four symbols, omitting their figure values, and

we

122

REFLECTIONS ON MAGIC SQUARES.

>,

here propose to indicate the order of o by

ri

ro by

@,

by

by +.

THE MAGIC SQUARE

In the introduction to the
find

IN CHINA.
Yih King, we

Chou

edition of the

scroll

some arithmetical diagrams and among them the Loh-Shu, the of the river Loh, which is a mathematical square from I to 9,
symbols, the

so written that all the

i.

e.,

yang

odd numbers are expressed by white dots, emblem of heaven, while the even numbers

THE SCROLL OF LOH.

Fig. 219.

THE MAP OF HO.*

CHINA.

(According to Ts'ai Y.uang-ting.)

TWO ARITHMETICAL DESIGNS OF ANCIENT

i.

are in black dots,

e.,

yin symbols, the

emblem of

earth.

The

in-

vention of the scroll

is

attributed to Fuh-Hi, the mythical founder

of Chinese civilization,

who according

to Chinese reports lived 2858-

2738 B. C.

But

it

goes without saying that

we have

to deal here

with a reconstruction of an ancient document, and not with the

document

itself.

The

scroll of

Loh

is

shown

in Fig. 219. in the

The

first

unequivocal appearance of the

is

Loh-Shu

form of

a magic square
*

in the latter part of the posterior

Chou dynasty

of Ho properly does not belong here, but we let it stand behelps to illustrate the spirit of the times when the scroll of Loh was composed in China. The map of Ho contains five groups of odd and even If the former are refigures, the numbers of heaven and earth respectively. garded as positive and the latter as negative, the difference of each group will uniformly yield -f- 5 or 5.

The map
it

cause

REFLECTIONS ON MAGIC SQUARES.

123

(951-1126 A. D.) or the beginning of the Southern Sung dynasty (1127-1333 A. D.). The Loh-Shu is incorporated in the writings
of Ts'ai

Yuan-Ting who

lived
I,

from 1135-1198 A. D.

(cf.

Mayers,

Chinese Reader's Manual,

754a), but similar arithmetical dia-

documents among scholars that lived under the reign of Sung Hwei-Tsung, which lasted from 1101-1125 A. D. (See Mayers, C. R. M., p. 57.)

grams are

traceable as reconstructions of primitive

The Yih King

its

is

unquestionably very ancient and the symbols

yang and yin as emblems of heaven and earth are inseparable from
contents.

They
in

existed at the time of Confucius (551-479 B. C.),

for he wrote several chapters

which are
(III,
I,

called appendices to the

Yih King, and

P-

them he says
i

IX, 49-50.

S. B. E.,

XVI,
4 9

3650

"To heaven belongs

to heaven, 5
;

to earth, 2

to heaven, 3

to earth,

to earth,

to heaven, 7

to earth, 8

to heaven,

to earth, 10.

to earth are five.

"The numbers belonging to heaven are five, and those belonging The numbers of these two series correspond to

each other, and each one has another that

mate.

may

be considered

its

The heavenly numbers amount

to 25,

and the earthly to

30.

The numbers of heaven and

earth together

amount

to 55.

It is

by these that the changes and transformations are spiritlike agencies kept in movement."
This passage was written about 500 B. C. and

effected,

and the

is

approximately

simultaneous with the philosophy of Pythagoras in the Occident,

who

declares

number
is

to be the essence of all things.

One

thing

sure, that the

cannot have been derived from Europe.

magic square among the Chinese It is highly probable, howand a general impulse

ever, that both countries received suggestions

from India and perhaps ultimately from Babylonia. But the development of the yang and yin symbols in their numerical and occult
China to a hoary antiquity so as to render it typically Chinese, and thus it seems strange that the same idea of the odd numbers as belonging to heaven and the even
significance can be traced back in

ones to earth appears in ancient Greece.

I

owe

the following communication to a personal letter from

124

REFLECTIONS ON MAGIC SQUARES.

Professor David Eugene Smith of the Teachers' College of

New

York:
"There
is

a Latin aphorism, probably as old as Pythagoras,

Virgil paraphrases this as follows
viii,
:

Deus imparibus numeris gaudet.

Numero deus impare

at

gaudet.

(Eel.

75).

In the edition

have

hand* there

is

a footnote which gives the ancient idea of the

nature of odd and even numbers, saying:

".
.

.impar numerus immortalis,

quiet dividi
licet

integer non potest,

par numerus mortalis, quia dividi potest;

goreos putare imparem
[a curious idea

Varro dicat Pythaideo medendi

numerum
I

habere finem, parent esse infinitum

;

which

have not seen elsewhere]

causa multarumque rerum impares numeros servari: nam, ut supra

dictum

est,

superi dii impari, inferi pari gaudent.

to the fact that the

"There are several references among the later commentators odd numbers are masculine, divine, heavenly,
I

while the even ones were feminine, mortal, earthly, but

at this writing place

cannot just

my

hands upon them.

at the

"As

to the

magic square, Professor Fujisawa,

Inter-

national Congress of Mathematicians at Paris in 1900,

assertion that the mathematics derived at an early time

made

the

from the

Chinese (independent of their

of a

own

native mathematics which

was

somewhat more

scientific

character), included the study of

first

these squares, going as far as the

400 numbers.
if

He

did not,

however, give the dates of these contributions,

indeed they are

known."

As
letter:

to other

magic squares, Professor Smith writes

found
in a

in

another

"The magic square

in the eleventh century.

is

It is also

work by Abraham ben Ezra found in Arabic works of the

twelfth century.

In 1904, Professor Schilling contributed to the

Mathematical Society of Gottingen the fact that Professor Kielhorn

had found a Jaina inscription of the twelfth or thirteenth century

* P. Virgilii Maronis
largyrii,
Pierii,
.
|
|

Opera,
. .
.
|

Accedunt
|

cum integris commentariis Scaligeri et Lindenbrogii

| |

Servii,
|

Phi-

Pancratius

Masvicius

Tom.

I,

Leonardiae,

. j

.
|

cloloccxvii.j

REFLECTIONS ON MAGIC SQUARES.

Khajuraho, India, a magic square of the notable peculiarity that each sub-square sums to 34." Fig. 220 is the square which Professor Smith encloses.
in the city of

We

must assume

that

we

are confronted in

many

cases with

an independent parallel development, but it appears that suggestions must have gone out over the whole world in most primitive times perhaps from Mesopotamia, the cradle of Babylonian civilization,
or later from India, the center of a most brilliant development of
scientific

and religious thought.

It

How old the magic square in China may be, is difficult to say. seems more than probable that its first appearance in the twelfth
is

century

not the time of

its

invention, but rather the date of a

Fig. 220.

recapitulation of former accomplishments, the exact date of which

can no longer be determined.

THE JAINA SQUARE.

Prof. Kielhorn's Jaina square
is

not a ".peefeet magic square"

according to
of
all

Mr. Andrews'

definition, epMtedrt&eme.

While the sums

the rows, horizontal, vertical, and diagonal, are equal, the

figures equidistant

sum

of the

first

from the center are not equal to n 2 and last numbers of the series. Yet it
is

I, viz.,

the

will

be seen

that in other respects this square

more
what might be
called absolute

a distribution of the figure values in

equilibrium.
First

we must
I

observe that the Jaina square

it

is

continuous,

by which

mean
itself

that

may

vertically as well as horizontally be

still

turned upon

and the rule

holds good that wherever

in

we

may

start four consecutive

numbers

whatever direction, back-

126

REFLECTIONS ON MAGIC SQUARES.

horizontal, vertical, or
34,

ward or forward, upward or downward, in slanting lines, always yield the same sum, viz.

which

is

2(n

-f-i)

and so does any small square of 2 X 2 cells. Since we can not bend the square upon itself at once in two directions, we make the result
visible in Fig. 221,

by extending the square

in

each direction by

half

its

own

size.

Wherever

4X4 cells

are taken out from this extended square,

we

shall find

them

satisfying all the conditions of this peculiar kind

of magic squares.

The

construction of this ancient Jaina equilibrium-square re-

quires another

method than we have suggested

for

Mr. Andrews'

10

REFLECTIONS ON MAGIC SQUARES.

inverted order.

127
in the first
is

We

do the same with the numbers

All that remains to be done

and

second vertical rows.

rest in

to

fill

out the

such a

way

as not to repeat either a letter or a number.

still

In
3,

the

first

row there are

missing for

C and D

the

numbers 2 and

of which 2 must belong to C, for

C3

appears already in the second

row and

is left

for D.
I

In the second row there are missing

and

4,

of which

must

belong to B, because
In the
first

we have B 4
row the

in the first
letters

row.
are missing, of

vertical

B and C

which

B must

1234

belong to 3, leaving

to 4.

In Consecutive Order.

The

Start for a Redistribution.

The Perfected

Redistribution.

Figure Values of the Square.

Fig. 222.

In the second vertical row

and

are missing for

and

2.

Aj and Do
then the
the

exist, so

must go

to 2,
all

and

to

i.

In the same simple fashion

the columns are

filled out,

and

cell names replaced by their figure values, which yields same kind of magic square as the one communicated by Prof. left

Smith, with these differences only, that ours starts in the

corner with

number

and the

vertical

rows are exchanged with

the horizontal ones.

ful

It is scarcely

necessary to point out the beauti-

symmetry

in the distribution of the figures

which becomes

fully

apparent

when we

consider their

cell

names.

Both the

letters,

A,

128

REFLECTIONS ON MAGIC SQUARES.

I,

B, C, D, and the figures,

2, 3, 4,

are harmoniously distributed

its dis-

over the whole square, so as to leave to each small square

tinct individuality, as

appears from Fig. 223.

Fig. 223.
i

The

center square in each case exhibits a cross relation, thus:

In a similar

way each one

of the four groups of four cells in

its

each of the corners possesses an arrangement of

symmetrically different from the others.

own which
p. c.

is

V-

A MATHEMATICAL STUDY OF MAGIC

SQUARES. A NEW ANALYSIS.
IV
/f"

AGIC

squares are not simple puzzles to be solved by the old

-*:*

rule of

"Try and

try again," but are visible results of "order"

as applied to numbers.

Their construction

is

therefore governed by

laws that are as fixed and immutable as the laws of geometry.

It will

be the object of this essay to investigate these laws, and

evolve certain rules therefrom.

have already been published by which various magic squares may be constructed, but they do not seem to cover the ground comprehensively, ft io the boliof

Many

rules

of-tiTC-writor-tlmt the rulQfi.hfrpin.mVrn will bn rnrnpctont

to.

pro

J?J

Fig. 224.

Fig. 225.

Fig. 226.

I3O

A MATHEMATICAL STUDY OF MAGIC SQUARES.

=b +g 2n = b + d 2c = d -\-m 2a = m + g
2k
It will

be seen that the

first

terms of these equations are the

quantities which occur

in the four corner cells,

is

and therefore that

the

the quantity in each corner cell

tities

mean between

two quan-

in the

two opposite
It is

cells that are

located in the middle of

the outside rows.

therefore evident that the least quantity in

cell

the magic square

must occupy a middle

it

in

one of the four

outside rows, and that

Since the middle

least quantity,
cell

cell

cannotjxcupy a corner celL of an outside row must be occupied by the

cells

and since any of these

of the upper

row by

rotating the square,

may be made the middle we may consider this

it is

cell to

be so occupied.
located the least quantity in the aqua***
plain

Having thus
corner
cells,

that the next higher quantity

must be placed
reflection in a
cells,
it

in

one of the lower

and since a simple

mirror would reverse

the position of the lower corner

smallest quantity

follows that the second

may occupy

either of these corner cells.

Next we

may

write

more equations
a

as follows:

+e

-f-

=S

(or summation)

d+e+g=S
h
also

+ e+ c=S

a+d+h=S
therefore

and

Hence the quantity in the central cell is an arithmetical mean between any two quantities with which it forms a straight row or
column.

A MATHEMATICAL STUDY OF MAGIC SQUARES.

13!

With
as

these facts in view a magic square

in Fig. 225.

may now

be constructed

shown

Let x, representing the least quantity, be placed in the middle

upper

cell,

and x -f

3;

in the

lower right-hand corner

cell,

y being

the increment over x.

Since

+y

is

the

mean between x and

v
in the

the quantity in the left-

hand central

cell, this cell

must evidently contain x

+ 2y.
cell,

Now
hand
cell

writing

lower left-hand corner

it

(con-

sidering v as the increment over x)

follows that the central right-

must contain x

+ zv.
cell in
filled

Next, as the quantity in the central

the square

is

mean

between x

2 y and

2z/, it

must be
cell

with

now

follows that the lower central

must contain

+ + y. It x zv + 2y,
v
-f-

and the upper left-hand corner cell x -\- 2.v -\- y, and finally the upper right-hand corner cell must contain x -{-v -\- 2y, thus completing the square which necessarily

must

frave

magic

qualifications
v,

with any conceivable values which

may

be assigned to x,

and

y.

and y which will produce a ^)(^ magic oqnnre rrmtniaing the numbers I to 9 inclusive in arithmetical progression. Evidently x must equal I, and

We
/

mav-aew proceed

to give values to x, v,

as there

must be a number
y=
i,

2, either

Assuming
suit,

if

=
-"

or

v or y must equal I also. 2, duplicate numbers would

3

re-

therefore v mih'
DLLciuoL
lliu

"T

lj1

'W

OK n^f

fe.

xg^ 7g4i
,r

>? .

luwlii LUitral cell is filled

with tho cymbolp

ap

ay,
e,i

and ac
i,

in this case thic

*^-

combination wurt equal 9; therefore,

if

them y

3,

or inco

verse*.

Using these

values, viz.,

I,

y =

and

v =

3,

the familiar

3X3

magic square shown

in Fig.

226

is

produced.
i

the series of numbers used has an

initial

number of

i,

n Fig. 226 and also a

constant increment of

i,

yot thio

may
is

bo conoidorod jK.only an

accidental feature pertaining to this particular square, the real fact

being that a magic square

of three

0/3X3
The

always composed of three sets

numbers .m^h.
is

difference between the

numbers of

each trio

uniform, but the difference between the last term of one

132

A MATHEMATICAL STUDY OF MAGIC SQUARES.

A MATHEMATICAL STUDY OF MAGIC SQUARES.

trio

133

and the

first

term of the next

trio-is

not necessarily the same as

trios.

the difference between the

numbers of the

will

For example, if x 2, y be as shown in Fig. 227.

=5

and v

= 8,
:

the resulting square

The

trios in this

square are as follows

2
10
18

7
15

12

20
28

23

The

difference between the

numbers of these

trios is
is

y
8.

5,

and the difference between the homologous numbers

=
is

A
in the

recognition of these*t*6o sets of increments

is essential to

the

proper understanding of the magic square. Their existence

masked

3X3

square shown

in Fig.

226 by the more or

less accidental

quality that in this particular square the difference

between adin Fig.

jacent numbers is always i. Nevertheless the square given 226 is really made up of three trios, as follows:
ist trio
i

2
5

2nd
3 rd
in

"
"

4
7

6 9
trios is

which the difference between the numbers of the

and the difference between the homologous numbers is 3. Furthermore it io pimply OH. acridentalr^&y. of this particular square thnt the diffemiro between the last term of a trio and tlicfirot

y v

I,

torm of tho next

trio

iff

T.

Having thus acquired a

clear conception of the structure of a

3X3
the

magic square,

we

are in a position to examine a

X9

com-

pound square

intelligently, this

square being only an expansion of

3X3

square, and governed by the same constructive rules. Referring to Fig. 229 the upper middle cells of the nine sub-

squares

may

first

be

filled^-m

thc-samc way that tho nine

t,

colic in

Fig. Jjij ^ei'o

filled,

using for this purpose the terms, x,

and

s.

Using
each

these as the initial terms of the subsquares the square

may
..

then be completed, using y as the increment between the terms of

trio,

the trios.

and v as thejncrement between the homologous_terms_of The roc(ihis shown in Fig. 228, in which the assignment of

Ov^fc^"

134

A MATHEMATICAL STUDY OF MAGIC SQUARES.

N
*x
X}

1
fc

N N N
a.

N
N
.

N
Xb

Xi

fc

Xi

Xi

N
Xi

x
^x
Xi

A MATHEMATICAL STUDY OF MAGIC SQUARES.

any values
square.
to x, y, v,
t

135

and

s,

will yield a perfect,

compound
^

9X9

Values

may

ass? be assigned to x, y, v,
I

and

which

will pro-

duce the series

with the

to 81 inclusive.

As

stated before in connection

I,

3X3

square,

x must

naturally equal

and

in order to

produce 2, one of the remaining symbols must equal I. In order to avoid duplicates, the next larger number must at least equal 3,

'1

and by the same prticggg the next must not be remaining one not less than 27. Because i-f- I
which
is

less

than 9 and the

::

+ 3 + 9 + 2 7=
is

4I

>

the middle

number of

the series

81, therefore just

these values
jli'iuiliuii

must be assigned
LliL

to the five symbols ucod in tho confixed,

I

uf
is

uqiuie.

The only symbol whose value

symbols

however,

x, the other four

may

have the values

9 or 27 assigned to them indiscriminately, thus producing

possible variations of a 9
If

all

the

is first

made

X 9 compound magic square. and y 2, and afterwards 3; is made

and

v =.
etc.

2,

the resulting squares will be simply reflections of each other,

may
is

Six fundamental forms of 9 9 compound magic squares be constructed as shown in Figs. 230, 231, and 232.

Only

six

forms

may

be made, because, excluding

trioa, if
tti

x whose

value

fixed^six different couples, (or

io

included)

may

be

made from the

It will

rest of the square

four^ymbols. becomes fixed.

Tbrm

r'cells being^eTelrmiried, the

be no\cd that theee are arranged in three grourjsoftwo

cquaroc each.on account of the curious fact that the squares in each
pair are

mutually convertible into each other by the following

process

If the

homologous

cells of

each

3X3

subsquare be taken in tbe>( 3 rqtnn?

if this

9X9 ihorofrom, a new magic 3X3 square will result.

order as they occur in the
is

square, pad a 3

ma4e

And

process

followed with
in

all

the cells and the resulting nine

3X3

squares are
will

arranged
result.

magic square order a

new

9X9

compound square

For example, referring to the upper square in Fig. 230, if the numbers in the central cells of the nine 3X3 subsquares are arranged
in

magic square order, the resulting square

will

be the

136
central

A MATHEMATICAL STUDY OF MAGIC SQUARES.

3X3

square in the lower

9X9

square in Fig. 230.

This

law holds good in each of the three groups of two squares (Figs. 230, 231 and 232) and no fundamental forms other than these can
be constructed.

The

question

may

be asked

How many

variations of 9

X9
may

compound magic squares can be made?

Since each subsquare

assume any of eight aspects without disturbing the general order of the complete square, and since there are six radically different, or
fundamental forms obtainable, the number of possible variations 6 89 *j
is

We *meyi^o\Vi.9QQ6e4r$&* &&&***
magic square as represented in Fig. 233.

the construction of a 4

X4

From our knowledge

A MATHEMATICAL STUDY OF MAGIC SQUARES.

3d.

137

Because the two middle terms of each of the two inside

columns (either horizontal or perpendicular) also compose the central square, their four end terms must likewise equal S.

2X2

We

may now ^writclhe

b
-\-

following equations:
-\-

c -}-v

=S

therefore

a
jfr

+ d = v + x,

which
cells is

shows^that

the

equal to the

sum of the terms in any two contiguous corner sum of the terms in the two middle cells in the

opposite outside column.

<3*kr

Because

g+h+n+m=S
and
it

follows that

or,^hat

the

sum

of the two end terms of any inside column, (either

horizontal or perpendicular) is equal to the

sum

of the two middle

terms in the other parallel column.

6tkr Since

t+o+n+d=S
and
h
therefore

+o+n+p

orjhe sum of
the

the two end terms of a diagonal

column

is

equal to

sum

of the two inside terms of the other diagonal column.

six laws -govern all
\

These

4X4

magic squares, but the s

ocillod^pcrfefft

4 squares also possess the additional feature that

the

sum

of the numbers in any two cells that are equally distant

in

from the center and symmetrically opposite to each other

square equals S/2. *^;n, n
i

the

r\

^n w

nirf *

^f^t

138

A MATHEMATICAL STUDY OF MAGIC SQUARES.

4X4
a
-\-

magiy
a^d
jp.

square.

corner eel/ we
Xj
-f-

Referring to^ig. 234, in the upper left-hand will placer a number ywhich may be represented by

by a

right-hand upper corner a number represented Also in tne central cells of the lower row we will write
in the

numbers represented respectively by a

lower left-hand
cell

3;

and a

t.

Then

in the

we

will place

and

in;

the central cells

a number represented by g x, of the outer right-hand column numbers

x and c x, and because the square represented respectively by b is to be perfect, we must write in the lower right-hand corner a number represented by

+ v.
4X4
square thus

The

unfinished perfect

made may now be

studied by the light of the laws previously given.

By

inspection-we see that

c

*\v=y+
We
also see that the central cells of the

occupied by the symbol

g together with
would

x and
result.

upper row should be v, by law 4, but if

thus occupied, duplicate numbers

It has,

however, been just shown that

:*^ = y +
and therefore g may be combined with y and t, thus producing diverse numbers, and still remaining correct in summation. Seeing that the square is to be perfect, the cell which is symmetrically opposite to that occupied

by a

+ y,

must be

filled

by a

number which

will

+ (g +

v), which

will

produce with a -f- y, a number equal to (a be g t, because

+ x)

/
In the same

*+=*+<
,

way the next cell to the left must be filled with g-\-y, and we may similarly fill the two inner cells of the left-hand
outer column with b
-f-

v and

+
all

v.

By
be
filled,

like

simple calculations

the remaining

empty

cells

may

thus completing the

4X4

square shown

in Fig. 235.

to the eight symbols used in Fig. 235 to

\A
e wfll

MATHEMATICAL STUDY OF MAGIC SQUARES.

now

proceed to show what numbers

may be

assigned

produce a perfect

4X4

magic Square containing the

rryfbes

to 16.

I/

is

evident that some pair

there/ore that one of ihe two symbols must equal

of symbols must equal I and i and the other

must equal
It
is

o,

(minus and fractional quantities being excluded).

evident that because a
-f-

also

=b

-f- c,

if

is

the

smallest

in the series, g must be the largest, and therefore numbers represented by a, b, c, g must form a series in which the means equal the extremes. In like manner x, y, t, v must

number

the four

also

form another similar

series.

1 6,

Supposing now that x and .since b and c are each

other,

and a

= o,

then g -f V must equal

less

than g, and must be also diverse

less

from each
that b

we
3,
i

find that

g Cannot be

than

3.

Supposing
evident

therefore that

g=
c,

then because a

+g=3=b+

c, it is

must equal

(or 2) and c must equal 2 (or i).

The

four

quantities a, b,

g may

therefore be assigned values as given below.

r b=i
g

/^
^\$ =
t t

--,

=3

f=i3equal 13 and y

As g-{-v=i6,--v must

By
this

inspection

it is

seen that either y or

must equal 14. must equal 5, and assigning

t

number

to y,

becomes

9,

or vice versa.

With

these values assigned to the symbols, Fig. 235 will Develop

the perfect

4X4

square shown Jn Fig. 236.

The

possible nurifi^eiNDf diverse

i

4X4 magic squares which

to 16 inclusive has

be constructed using; the numbers

been vari-

ously estimated by different writers, '880 changes having

toibre considered the

maximum number.

It

can howevj

proven that no

less

than 4352 of these squares may be^ tender the next heading.

14

A MATHEMATICAL STUDY OF MAGIC SQUARES.

STUDY OF THE POSSIBLE NUMBER OF VARIATIONS IN MAGIC SQUARES.

has been shown in connection with the
that there
bers,
is

magic square one possible arrangement of nine different numonly

magic square.
and
all

3X3

which

will constitute a

The
immense

4X4

larger squares

may however

be constructed

in great variety, their

number of

diverse forms increasing in an

ratio with every increase in the size of square.

Beginning with the

of the possible
the

4X4 square, in orftsr to solve the problem

variations that
it

number of

may

be constructed with

numbers

to 16 inclusive,

will

be necessary to consider the

relative properties of its

component elements, Which may be con'

veniently expressed as follows, although there

a're

several other sets

of eight numbers whose combinations will yield similar results.

=12

As

previously stated,

it

will be seen that

and

+g=b+c x+v=y+
a

f.

In consequence of this law

we

find that a

column

in a

X4

magic square m#y

pair of elements
pair, the latter

contain each of the eight qualities once (as in

the diagonal rows of square

shown

in Fig.

235).

In other cases a

may

be lacking, but be represented by another

outer vertical columns of Fig. 235).

being repeated in the column, (as shown in the two This ability to duplicate some

of the elements in place of others that are omitted leads to an

enormous amplification of the number of possible variations. If all the cells in any column are filled, (or any set of four cells, the summation of which is equal to a column) the remainder
of the square

may
rjpay

This Column

then be completed by the rules previously given. therefore be termed a "basic" row or column.
filled,

There are four plans by which a basic row may be

thus

A MATHEMATICAL STUDY OF MAGIC SQUARES.

14!
I,

making four III and IV.

classes of squares

which may be called Classes,

II,

\ For the sake of

brevity, the symbols a, b, c

x, y,
t

and g

will

be termed

the "a" elements,

and

and v the "x" elements.

Class I includes those squares in which the basic

row

is

made

up of

all

of the eight elements used once each.

Class II includes those squares in which one of the elements

used in the

first cell

of the basic

row

is

also used in $ie second

cell.

Fig. 237.

Fig. 238.

Fig. 239.

;, Class III includes those squares in v^iich an element of the

first cell in

the basic

row

is

also used in the third cell.

in.

Class
cell in

IV

includes those squares

which elements of the

and third
cells.

first

the basic

row are

also used in the second

Class I

may

be further divided into three Genera as follows

Genus

comprises those squares in which neither the outer nor

6.

inner pair of cells contain either

mean

or an extreme pair of "a"

or "x" elements. Fig. 237 represents a basic

row of Class

I,

Genus A.

a y

142

A MATHEMATICAL ST^DV OF MAGIC SQUARES.

outer cells contain a and g, and

tain b

x and

v,

and the two inner

cells

con-

and

c,

and

an^f y, as

shown

in Fig. 239.

Classes

II,

Ill/and IV have
aU',

but one genus each, and there are

consequently in

six different types.

specimens whicty each genus will yield, row into a complete square.

To we

determine the number of

will

now expand

a basic

Fi\ 240 shows

34X4
and
c

square in the-ripper row of which the

.

elements are written as previously given under Class

Filling the inner pair of cells in the lower

I,

Genus A.

row we

see that these cells

must contain a
c
-j- 3;

+v

-f-

y } but we have the choice of writing

in the right or left-hand cell.

is

the square

then completed by

the.

Choosing the right-hand cell laws previously given, and but

of each cell

slight attentioh is'srequired to

is

show that the contents

forced by these laws.

ele3,

This square will be magic for any values assigned to the ments, and it will be nt>rmaTif they are given the values I, 2,

and

o, 4, 8, 12.

To

find the

number of

possible squares of the above class

and

genus we

reflect that for the first cell

we have we
in

a choice of 16.

For

the fourth cell

we have
t

a choice of only
fiflst

4, since in the example,

having used a and y

either a, y,

in the

cell

are debarred from using

cells,

or

in the fourth cell.

Next, for the two central

we
the

evidently have a choice of 4,

and
fill

completing the square

we

have the choice of two methods to

the lower row.

Multiplying

number of

choices

16X4X4X2512.
and
sible
it is

therefore clear that Class

I,

Genus

will yield

512 pos-

forms of squares. Fig. 241 shows a square

of the lower row,

in

which the basic row of elements are

arranged so as to produce Class I

cells
it is

Genus B.

In

filling

the central

found that the equivalent of (a

3;)

(g

x} must be used, and there are three such

(1) (2) (3)

equivalents, viz.,

+x + (b + x
(a
) }

(g

+ y)

(& +.30

A MATHEMATICAL STUDY OF MAGIC SQUARES.

143

(3) ftowever will be found impossible, leaving only (i) and

(2)
available.

Choosing (i)

it

will

be seen that there are two

choices since n

+x

may

be located in either the right or left-hand

if

of the

two

cells.

Similarly

(2)

is

chosen, b

+x

may

be placed

in either of theseycells.
cell,

Hence
:

in, say,

the right-hand central lower

there

may

be placed

(1) a
(2) (3) (4)
as
is

+x g+y

shown

in Fig. 241, and',when

one of these four pairs of elements

It
is

used the remainder of the square becomes fixed.

first cell

therefore

follows that for the

16.

of the basic

row there
is

a choice of

For the fourth

cell

of same

row

there

a choice of 4.

For

the central cells of

same row therg

is

a choice of 4 and for the

lower row there

is

a choice of 4.

Multiplying these choices together

we have
which

/
the possible

16X4X4X4= 1024.
is

number

of variations of Class
I,

I,

Genus B.

Writing a basic

row of Class

Genus C,

as given in Fig. 239,

we
to

find that the equivalents of

fill

(a

+ x}

the central cells of the lower row.

-f- (g -\- v) must be used Because

a+g=b+c
and x

+v=

t -]-

there are no less than sixteen pairs which

to each other.

ma
v

be

made

all

equal

Ten

of these pairs however will be found unavailable,

leaving six pairs to choose from, and since each of these six pairs

may

be located in either of the two

cells,

there

is

a choice of 12

different

ways For the first

in

which the lower row may be

cell

filled.

of the basic row, there of the same

is

naturally a choice of
is

16.
cell

For the fourth

cell

row

there

no choice, as

this

must be

filled

with the complements of the

the basic

first cell.

For the

two middle

cells of

row

there

is

a choice of

4.

Multiplying

these choices together

we have:

144

A MATHEMATICAL STUDY OF MAGIC SQUARES.

16

X4X

12

= 768,
I,

which

is

the possible

number of

variations of Class

Genus C.
as

Proceeding
the lower

now

to Class II, a basic

row may be formed

n in Fig. 242.

It is

evident that neither a nor g can be used in

as equivalents of

row of the square, but

(a

+ v) +

(g
:

+ y)

we may

use either of the two couples

and since

either couple

a choice of 4 variations.
first cell

may be placed in either of two cells, To form the basic row, we have
as before.
first cell

there

is

for the
is

a choice of

F$r" the fourth cell there

choice of 6 seeing that one of the elements of the

.located therein, coupled with

must be
ele-

any One of the three remaining

Fig. 243.

ments of the opposite group.

choice of
2.

For the two inner

cells

there

is

Hence
16

for Class II

we have

X 6 X 2 X 4 = 768
row

varieties.

Class III has a basic

will

constituted as

shown

in Fig. 243.

It

basic

be found impossible to construct a magic square from the above row along the lines hitherto followed. Nevertheless, four

varieties of squares

may

be constructed on every basic row of this

class, on account of certain relations between the two groups of elements, which have not as yet been considered. The squares may

be

made

as herein before shown, but

when completed they appear

which
illustrate

to be, imperfect, as will be seen in Figs. 244-247,

These squares, although seemingly imperfect, are not actually so on account of a fuliar relationship between the numbers i, 2, 3, 4 and o, 4, 8, 12.
four /squares built up on the foregoing basic row.
Class III has for the
first cell

of the basic

row a choice of

16,

A MATHEMATICAL STUDY OF MAGIC SQUARES.

for the third cell a choice of 6, for the second cell a choice of 2, and

of each of the above forms there are 4 variations. 16

in this classk

Hence we have:

X 6 X 2 X 4 = 768
row
as

varieties

/
"has a basic

Class

IV

shown
be

in Fig. 248,

and the two

middle
couples

cells of the

lower row

may

filled

with either of the two

\
t)

tr;

Fig. 244.

Fig. 245.

Fig. 246.

Fig. 247.

thus permitting a choice of 4.

cell

of the basic

Having a choice of 16 for the first row, and a choice of 4 for the two inner cells of this
16

row,

we have

as a total:

X 4 X 4 = 256.

This square however has a peculiar property, owing to each

146

A MATHEMATICAL STUDY OF MAGIC SQUARES.

Summarizing the preceding

are in

results

it

will be seen that there

Class

T,

Genus

A MATHEMATICAL STUDY OF MAGIC SQUARES,

To
and
tP\e

complete the square, there are at least four available plans

resulting squares

may

be designated as Classes,

I, II,

III

and

IV

respectively.

writing the symbols of the "a" grcfup of columns across the square, in one direction, diagonal say from leftto right, and the symbols of the 'V grotfp also in
Clas\
I
is

made by

elements

ifk

diagonal columns, but in the opposite direction as shown in Fig. 251.

It will

be

\en

that the "a" elements occupy

Ijfe

right-hand

diagonal columns \and the "x" elements the lefVriand diagonal columns. Class I is only available when two elejhents in the end
cells

happen to be
Class
II.

ifj.

The Squares

in this class are

the elements in the basic-

row move by

a
JC

MATHEMATICAL STUDY OF MAGIC SQUA

different' sizes
cftv squares

that haveNbeen consider

increases

NOTES ON NUMBER SERIES USED IN THE CONSTRUCTION OF MAGIC SQUARES.

It

has long been

series

from

jf

magic squares may be constructed of numbers which do not progress in arithmetical

that
will

known

order.

Experiment

show, however, that any haphazard series

cannot be used for this purpose, but that a definite order of sequence
is

necessary which will entail certain relationships between different

the series.
'

members of

It will

therefore be our endeavor

a -v =s

-d

A MATHEMATICAL STUDY OF MAGIC SQUARES.

result.
It

therefore follows that the numbers composing a magic

square are not

made up

of a single group, but necessarily of

more

than one group.

Since
groups,

we have

eocn that the term "a" forms a part of two

write both groups as shown in Fig. 255, one horiand the other perpendicularly. zontally Next, by rule No. 5, it is shown that "the sum of the two end

we may

terms of any inside column (either horizontal or perpendicular)

equal to the

is

sum

of the two middle terms in the other parallel col-\-

umn."

It

therefore follows that v

=k+o

or v

b.

Using the term v as the initial number, we write this series perpenIn the same way it is seen that dicularly as shown in Fig. 256.
/

+g=n

-\- o,

or

=n

g, thus

forming the second hori-

zontal

column

in the

square (Fig. 257).

Next p

-\-

m=h

-{-

k or

Fig. 258.

Fig. 259.

Fig. 260.

Fig. 261.

=h
is

forming the third horizontal column and

in this

simple manner the square

It

may

be completed as shown in Fig. 258.

therefore evident that a 4

series of

X4

magic square may be

to permit

formed of any them

numbers whose

interrelations are such as

to be placed as
I

shown

in Fig. 258.

The numbers
struction

to 16

may
io

be so placed in a great variety of

ways, but the fact must not be lost sight of that, ao far ao the conof magic squarcc

concerned^ they only incidentally

possess the quality of being a single series in straight arithmetical

order, being really
in a

composed of as many groups as there are column of the square. Unless this fact is remembered, a ^^
mi
f-the series cannot be formed.

cells

clear

'

^>

conception of tho q

In illustration of the above remarks, three diagrams are given

in Figs. 259-261.

Figs. 259 and 260

show arrangements of

the

A MATHEMATICAL STUDY OF MAGIC SQUARES.

from which the diverse squares Figs. 262 and 263 are formed by the usual method of construction. Fig. 261 shows .1-lip nrrnnqrcmcnt o an irregular series of sixteen
I

numbers

to 16

numbers, which, when placed in the order of magnitude run as

follows
:

2-7-9-10-11-12-14-15-17-18-19-20-21-26-30-33

The magic square formed from this series is given in Fig. 264. In the study of these number series the natural question presents itself: Can as many diverse squares be formed from one series as
from another ?
This question opens up a wide and but
little

ex-

plored region as to the diverse constitution of magic squares. This idea can therefore be merely touched upon in the present article,

examples of several different plans of construction being given illustration and the field left at present to other explorers.
/

in

A MATHEMATICAL STUDY OF MAGIC SQUARES.

dicular

any

series of

yield

and diagonal all have the same summation, viz., 66. Hence numbers that can be arranged as shown in Fig. 258 will magic squares as outlined. But that it shall also produce

squares having the qualificationo that are termed "ptfilujB." may or may not be the case accordingly as the series may or may not be
capable of
still

further arrangement.
if

Referring to Fig. 254,

we amend our

definition

by now

call-

y
I
I

4 -///=/y-2
M
II II
II

29

/-3/=36-4
I i i i

46
22

Z/

*7

Fig. 266.

Fig. 267.

ing

it

a "poifoct" square,

we

at

once introduce the following

continuous equation:

we make our diagram of magic square producing numbers conform to these new requirements, the number of groups will at
and
if

once be greatly curtailed.

/o

38
2,4-

2?

8
/J"

//"*
/<f

zz

2/

/O
2.0

Z/

22

2S 28
J2 J?
Fig. 268.

/s

29
'7

29

Fig. 269.

The
diagram

multiplicity of algebraical signs necessary in our

is

amended
complete

so great that

it

can only be

studied in detail, the

diagram being a network of minus and equality signs.

The

result will therefore only be given here, formulated in the

all

following laws which apply in large measure to

I.

"perfect" squares.
series or

Perfect magic squares are

cells in

made

of as

many

groups

of numbers as there are

a column.

A MATHEMATICAL STUDY OF MAGIC SQUARES.

II.

Each

series or

group

is

composed of as many numbers as

there are groups.

III.

series
series.

two adjoining numbers of a must obtain between the corresponding numbers of all the
differences between any
initial

The

IV. The

do the second,

terms of the series compose another third, fourth terms and so on.

series, as

V. The differences between any adjoining numbers of these secondary series must also obtain between the corresponding terms
of
all

the secondary series.

illustrated

The foregoing rules may be square shown in Figs. 259 and

by the

series

and perfect

262.

Following and consequent upon the foregoing interrelations of these numbers is the remarkable quality possessed by the "perfect"

magic square producing

If the entire series

series as follows

is written out in the order of magnitude and between the adjacent numbers are written below, the differences the row of differences will be found to be geometrically arranged on each side of the center as will be seen in the following series

taken from Fig. 266.

3
-

4 -13-14-18-19-21-22-28-29-31-32-36-37-46-47
(6)

1914121
squares whether odd or even.

1214191

In the above example the number 6 occupies the center and the other numbers are arranged in geometrical order on each side of it.
It is the belief of the writer that this rule applies to all "perfect"

The

following example will suffice to illustrate the rule as

applied to a 5

magic square, Fig. 268 showing the

series

and

Fig. 269 the square.

1.4.7.8

.10.11.13.14.15.17.18.20.21.22.24.25.27.28.29.31.32.34.35.38.41

33I2I2I I2I2IJI2I2I I2I2I33

The diagram shown
it

in Fig.

270

is

given to impress upon the

the idea that a natural series of continuous numbers

may

be arranged in a great variety of different magic square producing be produced with any conseries. perfect

9X9

square^-tway-

A MATHEMATICAL STUDY OF MAGIC SQUARES.

ceivable values that

may
r^
1><

be assigned to the symbols

a, b, c,

d and

g,

used

in this

diagram.
c

symbols,

*^-tia^

qua

SLlLLlki-^jfthe ^nt-moi r l n 1^
is

numbers

i,

i, 3, 9,

27 for theseXj
It is

a must equal

i.

then

evident that for 2 there

a choice of four retts, as this

-\-

number may
-\-

be either a-\-b, a-\-

c,

d or a

-f-

g.

Selecting a

for 2,

makes
this

& =

i.

There
will

is

then a choice of three

-\-

oollo for 4,
3.

and for

number we

choose a

d,

making d

choice of

a
6

A MATHEMATICAL STUDY OF MAGIC SQUARES.

exactly the

same

plan,

and using the same breakmoves

the varia-

tions, radical as they

may

appear to be, are only so because different

series,
it

series of the

same numbers are employed, of which

has

been shown, there are at least twenty-four.

If the student will take Fig.

making "b"

(successively)

3, 9,

270 and fill in number values, and 27, he will acquire a clear

idea of the part taken in magic squares by the series conception.

The work
magic squares

o^f

determining the possible number of perfect

-fee-

9X9
of

may now

greatly simplified, for all elements are

i.

thus determined saving one,

progression.

e.,

the

number

of possible

modes

29
72 32
J/

20

6* S2 6s
6s
2<f

36 23 6 *
22 30 26

7s
77
76 /2

2S

V
60 "7

7*

79

46
/3

sc /o
66 S3

JO '7

62
Fig. 271.

may

be located in any of 80
variants

cells

and progress may be made

Theresquares will be at least

in

x ways, and 24

may

be constructed in each case.

fore, the possible

number
80

of different

9X9
1920,1-.

24

single

example

will serve to illustrate the possibilities

open

to x, the numerical value of

which

will

be

left for
let

the present for

others to determine.

As

previously given,

i i

b= d= g=

A MATHEMATICAL STUDY OF MAGIC SQUARES.

Then
Fig. 271
will

represent a 9

X9
i

square based on the

i
i

arrangement of symbols given in Fig. 270.

ii

Ii 1

f r*o n cf \T vl cY*r\ a, 1 1

order thesconstruction of this square must be governed by iHe fol-

lowing complicated rule:

Regular seeing: Three successive

diagonal.

cells in

upwa/d right-hand

Breakmoves
3 and

tween

4
7 10
13 etc.

Three

9
12

dowir and one to ( Extended'knight's mbve. )

cells

left.

and between

two

c^VIs to the right.

In
viz.
:

facjt,

however, th^ squar\

is

built

up by tie common
in

rule,

Regu
diagonal,
i

spacing :/Nine successive

all

cells

nd

br/akmoves, two Veils to the

in the follow
ig series:

upward right-hand rigr\t, the numbers

to 81 be

ng arranged

28.29.30

-7.8.
IninTim /n nr\T7A As shown/above,

34 35 36
the

55.56.57 58.59.60 61.62.63

i

etc.\

etc.

numbers

to8i may be aranged

in at

least

twenty-four

of such magic square producing series, thus giving

twenty-four Different squares, by the sarr% method of progression, and using tHe same breakmoves.
i

L. S. F.

CHAPTER

VI.

MAGIC SQUARES AND PYTHAGOREAN NUMBERS.

"I have compiled this discourse, which asks for your consideration and pardon not only because the matter itself is by no means easy to be handled, but also because the doctrines herein contained are somewhat contrary to those held by most of the Platonic philosophers." Plutarch.
'

rT^HE
-*-

mysterious relationships of numbers have attracted the

in all ages.

minds of men

The many-sided
scientific,

Franklin, whose 2OOth

literary

anniversary the philosophical,

and

worlds have

recently celebrated, used to amuse himself with the construction

of magic squares and in his memoirs has given an example of his

by showing a very complicated square with the comment that he believes the same to be tne most magical magic
skill in this direction,

square yet constructed by any magician.

That magic squares have had in centuries past a deeper meaning for the minds of men than that of simple mathematical curios we may infer from the celebrated picture by Albert Diirer entitled
"Melancolia," engraved in 1514.

The symbolism
listless

of this engraving

has interested to a marked degree almost every observer.

of the brooding genius sitting

The

figure

and dejected amid her unswaying balance, the flowbell,

completed labors, the scattered

tools, the

ing sands of the glass, and the magic square of 16 beneath the
these and other details reveal an attitude of

mind and a connection

of thought, which the great artist never expressed in words, but

left for

every beholder to interpret for himself.

discovery of the arrangement of numbers in the form of

The

magic diagrams was undoubtedly known to the ancient Egyptians

MAGIC SQUARES AND PYTHAGOREAN NUMBERS.

and
this

may have formed

part of the knowledge which Pythagoras

travels.

brought back from his foreign

ship of

We
into

have no direct evidence

that the Pythagorean philosophers in their studies of the relation-

numbers ever combined them

harmonic

figures, yet the

MELANCHOLY.
supposition that they did so
is

not at

all

improbable.

Such diagrams

and

their symbolic

meanings may

well have formed part of the

arcana of the esoteric school of Pythagoras, for similar facts were

accounted by ancient writers as constituting a part of the aporrheta

"f
MAGIC SQUARES AND PYTHAGOREAN NUMBERS.
of the order and the story
is

told of

an unworthy

disciple

who

re-

vealed the secret of the construction of the dodecahedron inscribed

within a sphere, this being a symbol of the universe.

Among

the best expositions of the Pythagorean philosophy are

and "Republic" of Plato. These dialogues were written after Plato's return from Magna Grsecia, where from contact with Archytas of Tarentum and other philosophers,
sections of the "Timseus"

he imbibed so

much

of the Italian school that his whole system of

ideas.
It is

philosophy became permeated with Pythagorean

even

lost writings

suggested that he incorporated into these dialogues parts of the of Philolaus, whose works he is known to have purchased.

No

portions

of the

dialogues

named have been more

commentators than the vague references to different numbers, such as the number 729, which is chosen to express the
puzzling to
difference between the kingly

man and

the tyrant, or the so-called

number of the
an endless

State in the "Republic," or the harmonic

number of
would be

the soul in the "Timaeus" of which Plutarch said that

toil

*'it

to recite the contentions

his interpreters-"
is

and disputes that have from

Either our text of these pas-

hence arisen
sages
is

among

corrupt or Plato

very obscure, throwing out indirect hints

only to those previously informed. Plato
"all writings are to

which would be

intelligible

states himself in the

"Phaedrus" that

be regarded

purely as a means of recollection for him

who

already knows," and

he, therefore, probably wrote more for the benefit of his hearers

than for distant posterity.

It is

upon the

principle of a

magic square that

wish to

inter-

pret the celebrated passage in the "Republic" referring to the

number

729, proceeding from

this to a discussion of certain other

numbers
efforts in

of peculiar significance in the Pythagorean system.

this direction are to
said,

My

be regarded as purely fanciful

the same

may

be

however, of the majority of other methods of interpretation.

The passage from

Socrates.

the "Republic" referred to (Book IX,

587-8,

Jowett's translation) reads as follows:

"And

if

a person

tells

the measure of the interval

in truth of pleasure,

which separates the king from the tyrant

he

MAGIC SQUARES AND PYTHAGOREAN NUMBERS.

will find him,

when

the multiplication

is

completed living 729 times

this

more

pleasantly,

and the tyrant more painfully by

calculation."

same

interval."

Glaucon.
Socrates.

"What a wonderful
life, if

"Yet a true calculation and a number which

closely

concerns

human

human

life is

concerned with days and nights

and months and years."

The number 729

is

found to be of great importance

all

through

the Pythagorean system.

Plutarch states that this was the number

belonging to the sun, just as 243 was ascribed to Venus, 81 to Mercury, 27 to the

opposite to ours).

moon, 9 to the earth, and 3 to Antichthon (the earth These and many similar numbers were derived
1 :2
:

from one of the progressions of the Tetractys,

:

14 :8

and

1 13

19 127.

The

figures of the above proportions were combined by

i, 2, 3, 4, 9,

Plato into one series,

in his

8, 27.

(Timaeus,
is

35).

Plutarch

"Procreation of the Soul," which

simply a

commentary

upon Plato's "Timaeus," has represented the numbers in the form

of a triangle; the interior
bers, 5, 13,

num-

and

35, representing

the

sums of the opposite pairs, were also of great importance.

The deep
thagoras

significance of the

Tetractys in the

system of Pyinferred from

Fig. 272.

may be

a fragment of an oath contained

in the

"Golden Verses."
Nat
fia

TOV

a/u-e'repov

faxf TrapaSovra rerpaKTov

Ilayav, devaov ^vcrews pt^w/u-ar' fyovo-av.

"Yea, by our Tetractys which giveth the soul the fount and source of ever flowing nature !"

Odd numbers were

especially

favored by the Pythagoreans

and of these certain ones such as 3 and its higher powers were considered to have a higher significance than others and in this way, perhaps, arose the distinction between expressible and inexpressible
or ineffable numbers
(dpifyioi farol KOL app^rot).

Numbers which

expressed some astronomical fact also held high places of honor,

MAGIC SQUARES AND PYTHAGOREAN NUMBERS.

as

may

be seen from a statement by Plutarch

(loc. cit.) in

reference
is

to the Tetractys.

"Now

the final
it is it

member

of the series, which

27, has this peculiarity, that

equal to the

sum

of the preceding

numbers

(1+2+3+4+9+8)
it

also represents the periodical

num-

ber of days in which the

moon

Pythagoreans have made

completes her monthly course; the the tone of all their harmonic intervals."

Fig. 273.

This passage indicates

the

sufficiently the

supreme importance of
the plan of a

number
If

27.

we

construct a magic square

27X27 upon
I

checker-board

arranging the numbers

to 729 first in numerical

order, then shifting the 9 largest squares

(9X9)

into the positions

indicated in the familiar

3X3

square, repeating the process with

MAGIC SQUARES AND PYTHAGOREAN NUMBERS.

the subdivisions of the
at the

-T ^T -

9X9

squares and so on

down

we

will arrive

1 following combination.

It will

be noted that

we have 365

white squares or days and

364 dark squares or nights

days."

a veritable "checkerboard of nights and

The number

365, the days of the solar year, very appro-

priately occupies the centre of the system.

zontals,

The columns,

hori-

and diagonals of the central square

3X3

foot

up 1095, or

the days of a 3 year period, those of the larger center square

foot

9X9
entire

up 3285 the days of a 9 year period, while those of the

combination

27X27

foot

up 9855^ the days of a 27 year period,

corresponding to the Tetractys

in other words, periods of years

i, 3, 9,

27.

We

may

with safety borrow the language of Plato and

''is

say that the above arrangement of numbers

concerned with days

and nights and months and years."

The
public"

interpretation of the other passage referred to in the "Re-

the finding of the

number of

the

State

(Book VIII,

546) has been a subject of the greatest speculation and by consulting the various editions of Plato it will be found that scarcely

any two
lation.

critics

3 agree upon a solution.

As Jowett

remarks,

it

is

a puzzle almost as great as that of the Beast in the Book of Reve-

Unfortunately
;

we have no
this

starting-point

from which to

and the very uncertain meanings of begin our calculations many of the Greek terms have caused many commentators to give up the solution of the problem
a hearer of Plato's, writes as
in sheer despair.
if

Aristotle,

who was

having a full knowledge of the mystery; Cicero, however, was unable to solve the riddle and his
sentiment became voiced in the proverb numeris Platonicis nihil
obscurius.

By
1

taking a hint from our magic square and starting with the

This method of constructing compound magic squares is, so far as I know, original with the writer. It bears some resemblance to the method of the numbers of each Schubert (see "Compound Magic Squares," p. 44) square, however, increase in periods of threes instead of by sequence. "Not only the perpendiculars, horizontals, and diagonals of this large square foot up 9855, but there are an almost indefinite number of zig-zag lines, which give the same footing.
;

3 Schleiermacher, Donaldson, and Schneider suggest 216, and much may be said in favor of this number. Jowett gives 8000 as the possible solution. Others suggest 951, 5040, 17,500, 1728, 10,000, etc.

MAGIC SQUARES AND PYTHAGOREAN NUMBERS.

number

27, I believe

we may
is

arrive at as

good a solution of the

problem as any that I have seen suggested.

pretation of the Greek terms
avt-foeu; 6vvdfj,evai re /cat

The following

inter-

offered.

the square of the

num27
2

dwaarevo/uEvai
pelq (nroGrdctLC,

ber times

its root,

X 1^27=

2187

increased by thrice the

first

terms

(of

the

Tetractys)
opovf ?ia(3ovoat
ofiotovvruv
TE

and four times whole series

of

the

(i_|_2+3+4+9-f-84-27)X4= 216

Kal

dvo-

potowTuv

Kal dvgovTuv

numbers unlike yet bearing the same ratio

Kal (j>6tv6TVu)v

whether increas-

ing or decreasing
(i.

e. i :2

14 :8

or 8:4:

:2 :i It

may

also refer

to the

ascending and descending figures

8, 4, 2, i, 3, 9,

of the triangle.

27)

Trdvra irpoarjyopa Kal

irpbs
d/l/l?7/la

prjTa.

anetyrivav

makes the sum commensurable and expressible in

parts.
(i,

all

its

sum=
i, 2, 3, 4, 5,

2460

e.

2460 is easily divisible by

6, 10, 12

etc.)

this

sum increased by

A
1

2460X1^=
5

3280
3285

ovvyei<;
rpi? avfyOeic

and adding
is

3280+5=
32 8 5X3==

multiplied by 3

This solution of the problem, 9855, it will be noted, brings us again but by a different route to the magic number of our large
square.

The second

part of the passage contains a description of the above calculation

the

number by which

may

be verified.

6vo dp/uovias Trapex f:Tai

(the

number) yields two harmonic parts,

of

rfjv fiev ICTJV Jtrd/af,

one

which

is

square
kKa.TbvTooavTdK.is,
Tr/v 6e laoprjKT)

3X3=

multiplied by too:
*

9X100==

900

pev,

h e other has one side

equal to the square
3

rr?

Trpo^KEi

6e,

and the other oblong

3X2985=8955

sum=

9855

MAGIC SQUARES AND PYTHAGOREAN NUMBERS.

63

The remainder

of the passage describes the length of the ob'

:

long which we have shown above to be 2985

K.arbv
fjLEv

dpt6fj.(Jv

CLTTO

(the

6ia/j.Tpuv Tre/zTrddof,

100 oblong) is times the side of a

rectangle having di-

agonals of
(i.

5.
3

100X3=
and
4.)

e.

having sides of

ojfruv Seofjiivuv ivbq indc-

less of

one each of the

i.

ruv,

expressible parts,
4

e.

and

appTjTuv 6e dveiv,

and

2 of the inexpressi-

300

(5+4+3~f~3)^

285

ble
EK.O.TOV

6e Kvfiuv rpiddos

plus 100 times the cube

3

of 3

(3)

Xioo=

2700
2985

sum=
Plato states that the
metrical figure
r

number of

the State "represents a geo-

w hich has control over the good and evil of births. For when your guardians are ignorant of the right seasons and unite bride and bridegroom out of due time, the children will not be
goodly and happy."

The number

9855, expressing a period of

27 years, might thus represent the dividing line between the ages

when men and women should begin to bear children to the State, 20-27 years for women, 27-34 years for men. ( See also "Republic," Book V, 460). Aristotle in his "Politics" (V, 12. 8) says in
reference to the

number

of the State that

is

when

the progression of

number
the

is

increased by 1 / 3 and 5

added, 2 harmonies are produced

seen from our analysis of
reference to the

giving a solid diagram.

first

This, as

part of the passage,

may be may have

2

number

3285, which, being represented by 3 X36s,

may

be said to have the

dimensions of a

solid.

In his "Reflections on Magic Squares" Dr. Carus gives some

very striking examples of the relationship between magic squares

and the musical figures of Chladni.

tions

would

like to

touch before

concluding upon a closely related subject and

show

certain connec-

which exist between the magic square, which we have constructed, and the numbers of the Pythagorean harmonic scale. This
scale had,

however, more than a musical significance

among

the

MAGIC SQUARES AND PYTHAGOREAN NUMBERS.

Greek philosophers;
in
it

was extended

to

of planetary movements and above

all else

comprehend the harmony to represent the manner

It is especially

which the "soul of the universe" was composed.

in the latter sense that Plato

employs the scale in his "Timaeus."

In a treatise by Timseus the Locrian upon the "Soul of the

World and Nature," we

find the following passage

"Now

all

these

proportions are combined harmonically according to numbers, which proportions the demiurge has divided according to a scale scientifically,

so that a person
is

is

not ignorant of what things and by what

means

combined; which the deity has not ranked after but he made it older by taking the the substance of the body
the soul
,

first
it

of unities which

is

384.

Now

of these the

first
all

being assumed

is

easy to reckon the double and triple; and

the terms, with

their
lation

complements and eights must amount to 114,695."

(Trans-

by Burge.)
is

Plato's account of the combination of the soul

to the above,
first

very similar

though he seems to have selected 192, (384/2) for the number. Plutarch in his commentary makes no mention of
first
2

Timseus, but states that Grantor* was the reason that

it

to select 384, for the

is

represented the product of 8

X6, and

the lowest

number which can be taken

leaving fractions.

seen mentioned, a

is

by eighths without Another very possible reason, which I have not 2 that 384 is the harmonic ratio of 27 /2 or 364.5,
for the increase
closely the days of the year.
:

number which expresses very

243 1256 1364.5

1384.

The proportion 243:256(3 5 :44 ) was employed by the Pyth5 agoreans to mark the ratio which two unequal semitones of the
harmonic scale bear to one another.
Batteux has calculated the 36 terms of the Pythagorean scale starting with 384 and his series must be considered correct, for it
fulfils

the conditions specified by Timseus,

the

numbers

all

footing
treatise

Grantor lived nearly 100 years after Timaeus the Locrian.

The

upon the "Soul of the World and Nature," which bears the
probably belongs to a
5

latter's

name
and

much

later period.
36,

For further references to this ratio see Plato's "Timseus," 18. Plutarch's "Procreation of the Soul,"

MAGIC SQUARES AND PYTHAGOREAN NUMBERS.

up
1

14,695

few of the numbers of

this

harmonic

scale

marking

the "first unity" and several of the semitones will be given.

f

E
C F

384

ist

octave
-j

486

[
,

729

(For

Batteux's full series

of
is

and
the

method

calculation

2nd octave

F
c
!944
.

reader

referred to Burge's translation of Plato Vol. VI.

3 rd

octave

j
(

p. 171).

flat

4th octave

B-flat 4374
it

By

referring to our magic square

will be noted that the

-first

of unities," 384, constitutes the magic

number of
If

the small

3X3

square beginning with the

number
series of

100.

we arrange

the magic

numbers of the 81 squares (3X3)

in the order of their

we

find that they fall into

magnitudes numbers, each series beginning 9

as follows:
I

II

III

IV
816

V
1059

VI
1302

VII
1545

VIII
1788

IX
2031

87

330

573

The

intervals

between these

series are

worthy of note.

INTERVALS.
Between
I I
I I
I

and
"
" " "

II

243

the

first

member

of the ratio 243:256.

III

486

IV

V
VII

729 972
1458 1944

"

IX

C F C F c

of the ist

"
"

" " " "

octave " ist " 2nd

"
"

2nd
3 rd

" "

If in the

we arrange
same way,

the magic numbers of the large squares

it

(9X9)

will be

found that they

fall into

3 series of 3

numbers, each series beginning

I

n
3204
I

in
5391
B-flat of the 3rd octave. " " B-flat " 4th

1017
Interval between

and

II

" 111

= 2187 = 4374

Numerous other
bers, but these

instances might be given of the very intimate

connection between magic squares and various Pythagorean

numin

must be

left for the

curious-minded to develop for

themselves.

Such connections as we have noted are no doubt

MAGIC SQUARES AND PYTHAGOREAN NUMBERS.

respects purely accidental, being due to the intrinsic harmony numbers and therefore not implying a knowledge by the ancients of

some

of magic squares as

we now know

them.

The harmonic arrangement

forms both plane and

by the Greeks of numbers

solid

in geometrical

may, however, be accepted, and Plato's descriptions of various

numbers obscure and meaningless as they were to succeeding generations, may have been easily comprehended by his hearers when
illustrated
6 by a mathematical diagram or model. Differences between the methods of notation

in ancient

and

modern times have

Greeks by

necessarily produced differences in the concep-

tion of numerical relations.

letters of the

The expression of numbers among the alphabet was what led to the idea that every
attribute,

name must have a numerical

letters

but the connection of the

of the

name was

in

many

cases lost, the

itself.

number being

re-

garded as a pure attribute of the object

similar confusion

of symbols arose in the representation of various concepts by geometrical forms, such as the five letters of
tion of health

YrEIA and

the symboliza-

by the Pythagoreans under the form of the pentalpha

star.

or five-pointed
It

was

the great defect of the Greek schools that in their search

for truth, methods of experimental research were not cultivated.

Plato in his "Republic" (Book VII,

piricists,

530-531) ridicules the em-

who sought knowledge by

studying the stars or by comis

paring the sounds of musical strings, and insists that no value

to be placed

upon the testimony of the

is

senses.

"Let the heavens

alone and train the intellect"

If the

his constant advice.

medes

in statics

examples set by Pythagoras in acoustics and by Archihad been generally followed by the Greek philos-

ophers, our knowledge of natural phenomena might have been advanced a thousand years. But as it happened there came to prevail but one idea intensified by both Plato and Aristotle, and handed

down through
8

the scholastics even to the present time, that knowl-

The description of the number of the State in the "Republic" and that of the Soul in the "Timseus" render such a mode of representation almost necessary. Plutarch ("Procreation of Soul," 12) gives an illustration of an harmonic diagram 5X7 containing 35 small squares "which comprehends in its subdivisions all the proportions of the first concords of music."

MAGIC SQUARES AND PYTHAGOREAN NUMBERS.

edge was to be sought for only from within.

Hence came the

flood

of idle speculations which characterized the later Pythagorean and Platonic schools and which eventually undermined the structure of
ancient philosophy.

But beneath the abstractions of these schools

one can discover a strong undercurrent of truth. Many Pythagoreans understood by number that which is now termed natural law.

"Number
As

Such undoubtedly was the meaning of Philolaus when he wrote is the bond of the eternal continuance of things," a sentiphysicist could not express

ment which the modern

the
first

more

fittingly.

study of importance for the youth of his "Republic"

Plato selected the science of numbers; he chose as the second ge-

ometry and as the third astronomy, but the point which he emphasized above all was that these and all other sciences should be
studied in their "mutual relationships that

we may

learn the nature

states, "will

of the bond which unites them."

"For only then," he

be well bestowed."

a pursuit of them have a value for our object, and the labor, which

might otherwise prove

ance! and

fruitless,

Noble

utter-

greater need of this at the present day with our complexity of sciences and tendency towards narrow specialization.

how much

In the

spirit

of the great master

whom we

have just quoted

we may compare

the physical universe to an

Isolated investigators in different

immense magic square. areas have discovered here and

no regard to the

there a few seemingly restricted laws, and paying

territory

beyond their confines, are as yet oblivious of the great pervading and unifying Bond which connects the scattered parts
and binds them into one harmonious system. Omar, the astronomer-poet, may have had such a thought in mind, when he wrote:
"Yes and a single Alif were the clue Could you but find it to the treasure-house And peradventure to the Master too;
;

Whose

secret presence, through creation's veins

Running

quicksilverlike eludes your pains;" etc.

is

When
clue,

Plato's advice

followed and the "mutual relationships

between our sciences" are understood we

and having found

it

may perchance find this be surprised to discover as great a sim-

tsrf
MAGIC SQUARES AND PYTHAGOREAN NUMBERS.
plicity

underlying the whole fabric of natural phenomena as exists

in the construction of a

magic square.
c. A. B.

MR. BROWNE'S

SQUARE AND LUSUS NUMERORUM.

Jr.
is

The 27

27 square of Mr. C. A. Browne,

its

interesting

because, in additon to

arithmetical qualities

commonly possessed
as well as the
first

by magic squares,

it

represents

some

ulterior significance of our

calendar system referring to the days of the

month

days of the year and cycles of years.

sight mystifying, to observe

It is

wonderful, and at

how

the course of nature reflects even

relations
;

to intricate details the intrinsic

harmony of mathematical

and yet when we consider that nature and pure thought are simply the result of conditions first laid down and then consistently carried out in definite functions of a distinct and stable character, we will

no longer be puzzled but understand why science is possible, why man's reason contains the clue to many problems of nature and,
generally
speaking,

why

reason with

all

its

wealth of a priori
sight seems to be

thoughts can develop at

all in

a world that at

first

a mere chaos of particular facts.

The

purely formal relations of

mathematics, materially considered mere nonentities, constitute the

bond of union which encompasses the universe,

motes, the motions of the Milky
binations of chemical

stars as well as

Way

not less than the minute com-

atoms, and also the construction of pure

thought in man's mind. Mr. Browne's square

cause
it

is

of great interest to Greek scholars be-

throws light on an obscure passage in Plato's Republic, referring to a magic square the center of which is 365, the number of

days in a year.

The

construction of Mr. Browne's square

is

is

based upon the

simplest square of odd numbers which

3X3-

But

it

becomes

somewhat complicated by being extended

which
is

to three in the third

power
up

27.

Odd magic
in

squares, as

we have

seen, are built

by a progression
that
fall

staircase

fashion,

but since those numbers

outside the square have to be transferred to their cor-

MAGIC SQUARES AND PYTHAGOREAN NUMBERS.

responding places inside, the
into the knight's
first

**%<

and

last staircases

are changed

move

of the chessboard, and only the middle one

retains

its

original staircase form.

We

so that the central figure, which in a

fall

must construct the square 3X3 square is 5, must always

in

the central

cell.
I

Accordingly,

we must

start

the square
cell

beginning with figure

outside of the square in any middle

it,

immediately bordering upon

from which we may

either proceed

which gives four starting-points from the right or the left, either

upwards or downwards which yields eight possibilities of the 3X3 For the construction of his 27X27 square, Mr. Browne square.
might have taken any of these eight
possibilities as his

pattern.

THE EIGHT POSSIBLE ARRANGEMENTS OF THE 3X3 MAGIC SQUARE.

Fig. 274.

He

selected the one starting

right,

on the top of the square and moving

It is

toward the

and thus he always follows the peculiar arrangethe fourth of the eight arrangewill take the trouble to will find that
it

ment of

this particular square. in Fig. 274.

ments shown

Any one who

trace the regular succession of

is

Mr. Browne's square

a constant repetition of the knight's move, the staircase

move

and again a knight's move on a small scale of 3X3 which is repeated on a larger scale 9X9? thus leading to the wonderful regularity which, according to Mr. Browne's interpretation of Plato, astonished
the sages of ancient Greece.

Any one who

discovers at

random some magic square with

its

ft*
MAGIC SQUARES AND PYTHAGOREAN NUMBERS.

immanent harmony of numbers, is naturally impressed by its apparent occult power, and so it happens that they were deemed supernatural and have been called "magic."

They seem

to be the product
ulterior

of some

secret intelligence

and to contain a message of

necessary and intrinsic.

meaning.

But

if

we have

the key to their regularity

is

we know

that

the harmony that pervades them

Nor is the regularity limited to magic squares. There are other number combinations which exhibit surprising qualities, and
I will

here select a few striking cases.

ing order
ber

write down all the nine figures in ascending and descendwe have a number which is equal to the square of a numconsisting of the figure 9 repeated 9 times, divided by the sum
If

we

of an ascending and descending series of

12345678987654321

all

the figures thus

999999999X999999999

I+2+3+4+5+6+7+ 8+ 9 +8+ 7 +6+5+4+3+2+i*

is

The

secret of this mysterious coincidence

;

that

11X11=121;
I

111X111=12321

1111X1111=1234321,
series

etc.,

and a sum of an
with
is

ascending and descending

equal to the square of
its

which

starts

always

3+4+3+2+1=4X4,
instance of the

highest number. etc., which we will

1+2+1=2X2; 1+2+
illustrate

by one more

same

kind, as follows:

7777777X7777777
i

+ 2 + 3+4+5+6+7+6+5+4+3 + 2 +
regularities.

There are more instances of numerical

All

numbers consisting of
list:

six equal figures are divisible

by

7,

and

also, as a matter of course,

by 3 and n, as indicated

in the

following

111111:7=15873 222222:7=31746 333333 7=476i9 444444:7=63492

:

666666:7=95238
777777:7^=111111

888888:7=126984 999999:7=142857

MAGIC SQUARES AND PYTHAGOREAN NUMBERS.

Finally

we

will offer

two more strange coincidences of a lusus

numerorum.

0X9+1=1
I2X9+3 =I11 123X9+4=1111

123456X9+7= * 1 1 1234567X9+8=1 1 1 1 1 1 1 1 12345678X9+9=1 1 1 1 1 1 1 1 1 123456789X9+ 10=1 1 1 1 1 1 1 1 1 1.

1X8+1=9
12X8+2=98 123X8+3=987
1234X8+4=9876 12345X8+5=98765 123456X8+6=987654
1234567X8+7=9876543 12345678X8+8=98765432 123456789X8+9=987654321.

No wonder
mind.

that such strange regularities impress the

human

A man

who knows only

is

the externality of these results will

naturally be inclined toward occultism.

The world

of numbers as

much

as the actual universe

full

of regularities which can be

reduced to definite rules and laws giving us a key that will unlock
their mysteries
nite conditions.

and enable us to predict certain

results

under

defi-

Here
is

is

the key to the significance of the a priori.

its

Mathematics

a purely mental construction, but

compo-

sition is not arbitrary.

On

the contrary

it

is

tracing the results of

our

own

doings and taking the consequences of the conditions

we
its

have created.
possibilities

Though

the scope of our imagination with

all

be

infinite, t'he results

of our construction are definitely

determined as soon as

we have

laid their foundation,

and the actual

world

is

simply one realization of the

infinite potentialities of being.

Its regularities

can be unraveled as surely as the harmonic relations

of a magic square.

MAGIC SQUARES AND PYTHAGOREAN NUMBERS.

Facts are just as

can but gain a clue

their nature,

much determined as our thoughts, and if we to their formation we can solve the problem of

and are enabled to predict their occurrence and sometimes even to adapt them to our own needs and purposes.

study of magic squares

may have no

practical application,
if
it

but an acquaintance with them will certainly prove useful,

were merely to gain an insight into the fabric of regularities of any

kind.
p. c.

(SOME CURIOUS MAGIC SQUARES AND\:OMBINATIONS.

TV curious and interesting magic squares and combinations <*** have been devised by the ingenious, a selection of which will

TAN Y

be given in the following pages, some of the examples being unrepresented for the
first

time in print.

__

one corner diagonal is composed of. odd numbers and the other corner diagonal of even numbers. It is stated that no square can be made which has a perrig. 275 possesses the unusual feature that

/2

/o

burs.

In this square the numbers

to

16 are arranged in symline.

I

metrical order in relation to the heavy center

Fig. 276
is

peculiar in

having the
I

numbers

and 2

in the

same

row, and Fig. 277 contains

combinations being unusual.

and 3

in the

same row, both of these

The

curious irregularities of tto 6

first

ferred to in the

X -6 magic squares were reand many unsuccessful attempts have chapter,

this order.

been made to construct regular squares of

novel and

174

SOME CURIOUS MAGIC SQUARES AND /COMBINATIONS.

interesting

6X6

square

is

illustrated

work

entitled

Games,

Ancient and Oriental by Geo. Falkene/,* and is here reproduced in It will be seen however/that the two corner diagonals Fig. 278.
of this square do not the

sum HI,

r^ut

by a jtransposition of some of
corrected, an

figure^

this imperfection

shown

in Fig. 279.

Other transpositions are also possible which

/6

will effect the

same

SOME CURIOUS MAGIC SQUARES AND COMBINATIONS.

3.

175

The

sixteen

2X2

sub-squares have a constant summation

of 130.
4.

Each quarter contains four

of which

3X3

squares the corner numbers

sum

130.
is

5.

Any
has

5X5
its

square which

contained within the

8X8 square
same

corner numbers in arithmetical sequence.

is

very interesting class of squares

pp. 337-338

referred to in the

work on

and 339 as follows:

''The Rev. A. H. Frost, while a missionary for

India, of the

many

years in

Church Missionary

Society, interested himself in his

76

SOME CURIOUS MAGIC SQUARES AND COMBINATIONS.

adepts in the formation of such squares from time immemorial, not

only does Mr. Frost give his an Indian name, aad Mr. KooBon^ywci
hia ^Cfti9Btt"a foealitu
1*11

Eaotorn Eui'opc, not only

is

one of these

squares represented over the gate of Gwalior, while the natives of

India wear them as amulets, but

La Loubere, who wrote

is

in

1693,

expressly calls them .'Indian Squares.'

"In these Indian squares

it

necessary not merely that the

summation of the rows, columns and diagonals should be alike, but that the numbers of such squares should be so harmoniously balanced that the summation of any eight numbers in one direction as in the moves of a bishop or a knight should also fre alike."

An
slight

example of one of these squares

it

is

given in Fig. 282 and

examination will show

to be of the

same order as the

6
52

'7

J2

JO

5
20

/o

J/

29
Fig. 283.

26

2*
Fig. 284.

"Jaina" square described by Dr. Carus in a previous chapter (p.

125
ff.),

but having enlarged characteristics consequent on

It will

its

in-

crease in size.

be seen

-that

the extraordinary properties as

quoted above
cell in

in italics exist in this square, so that starting

from any

the square, any eight

numbers that are covered by eight con-

secutive similar

moves

will

sum

260.

In addition to this the

num-

bers in every

2X2
I

sum 130

thus,
i -\-

square, whether taken actually or constructively, 16 61 16 130 and I 52 -f 58 55

130, also

58

+4

+ + +3
1

etc

Furthermore, (as in the

Jaina square) the properties of this square will necessarily remain

unchanged

if

columns are taken from one side and put on the other,

SOME CURIOUS MAGIC SQUARES AND COMBINATIONS.

or
if

177
it

they are removed from the top to the bottom, or vice versa,

-being a perfectly continuous square in every direction.

The wonderful symmetry

tion to the

of this square naturally invites atten-

method of

its

construction,

which

is

very simple, as
I

may

be seen by following the natural sequence of the numbers

in Fig.

to 32

half of the series.

283 which shows the disposition of the numbers of the first The second half is simply a constructive repeti-

tion of the first half.

The numbers
similarly

of this square are arranged

located cells
i

symmetrically

in

relation to

in

diagonally

opposite quarters, thus, (referring to Fig. 282)

+ 64 = 65

and

61

=65

etc.

This feature permits the completion of Fig. 283

by

filling in

the vacant cells at

random with

their respective differ-

ences between 65 and the various numbers already entered.

square constructed by the same method Fig. 284 shows and having similar properties, with natural limitations due to its
small
size.

34X4

This square strikingly resembles the Jaina square as

(See Fig. 222, p. 127) the numbers and arrangement of same in the two corner diagonal columns being identical in both squares, while the other numbers are differently
modified by Dr. Carus,
located.

an original square contributed by Mr. L. S. Frierson, which combines to a limited extent some of the curious
Fig. 285
is

8X8

characteristics of the Franklin

and the Jaina or Indian squares.

:

It

possesses the following properties

1.

Considered as a whole

it is

an 8

X8

magic square.

2.
3.

Each quarter

magic square. The four central horizontal columns make two

squares.

is

in itself a

4X4 magic

4. It contains twenty-five

2X2

squares, each having a con-

stant
5.

summation of

130.

It also

contains twenty-four

3X3

squares, the four corner

cells
6. 7.

of which have a constant summation of 130.

Any
In

4X4 square has a constant summation of 520. any 5X5 square the four corner cells contain numbers
rectangular parallelogram which
is

in arithmetical sequence.
8.

Any

concentric with

178

SOME CURIOUS MAGIC SQUARES AND COMBINATIONS.

any of the nine sub-centers contains numbers
cells that will

in its corner

sum

130,

excepting when

the diagonals of
side of the

any of the four sub-squares of 4

parallelogram.
9.

X4

form one

Any

octagon of two

cells

on a

side, that is concentric

with

any of the nine sub-centers will have a constant summation

of 260.
10.

No
may

less

than 192 columns of eight consecutive numbers

be found having the constant summation of 260 as

:

follows (See Fig. 286)

/\

/\

>

y /\

X
Fig. 285.
Fig. 286.

Horizontal columns

8 8
(

Perpendicular columns Perpendicular zig-zags

Horizontal zig-zags

to

Ai )

(A

to

A2)
to

8
2

Corner diagonals
Constructive diagonals

(D

Di )

Bent diagonals (as

in

Franklin squares) (T to T\ and

16

to

To)

Columns

partly straight and partly zig-zag (as V toV\) 88 Columns partly diagonal and partly zig-zag (as Pto Dj) 32

Double bent diagonal columns (as

to

N)

16

Total

192

SOME CURIOUS MAGIC SQUARES AND COMBINATIONS.

179

Mr

Frierson has also constructed an 8

X8

square shown in
it

Fig. 287,

which

is

still

more curious than {he

last one, in that

perfectly combines the

salient features of the Franklin

and the In-

dian squares,

viz.,

the bent and the continuous diagonals, besides

6o 5 9 y x^y
,
,

JTJ

sz
s/

i^^y

66

62
/6
66

/o

60

ss

SO

39 /O
29
66

60
2-4

33

42
4/

4-7

60

39

46 26
Fig. 287.

2O\4S 66
Fig. 288.

29\Z8\4O yv y

exhibiting

many

other interesting properties,

:

some of which may

be mentioned as follows
1.

Any

2 square has a constant summation of 130,

cells

2.

The corner The corner

The corner
to

of any

3X3

square which

lies

wholly to

the right or left of the axis

3.

cells

of any

AB sum 130. 2X4, 2X6 or 2X8

rectangle

perpendicular to
4.

AB

cells

and symmetrical therewith sum 130. of any 2 X 7 or 3 X 6 rectangle diagonal

12

AB sum

130, as

19 +46 =
5.

+ 50 + 45 + 23 =
5X5

130,

49

+ 16 +

130

etc., etc.

The corner

cells

of any

square contain numbers in

arithmetical progression.
6.
7.

8.

Any Any Any

constructive diagonal column

sums 260.

bent diagonal sums 260.

reflected diagonal

sums

260.
in dotted lines

(NOTE: Reflected diagonals are shown

on Fig. 287.)

By

dividing this square into quarters, and subdividing each

quarter into four

2X2

squares, the

numbers

will

be found sym-

i8o

SOME CURIOUS MAGIC SQUARES AND COMBINATIONS.

metrically arranged in relation to cells that are similarly located in

diagonally opposite
65, 57 + 8 = 65
etc.

2X2

squares in each quarter, thus

64

+ =
i

2TL

36

23 24

fS ss
/o
Af

/f

JY /J

/z
3/5

60

JTJ

zo

26

60

/Z 22
to

27 2S
29

fS

J/
'7

J/
Fig. 289.

JJ
Fig. 290.

2.4-

49

Another
which
its

8X8

square by Mr. Frierson

for
its

is

given in Fig. 288

is

alike remarkable

constructive simplicity and for

curious properties.

principal features of the

Like Fig. 287 this square combines the Indian and the Franklin squares in its

bent and continuous diagonal columns.

SOME CURIOUS MAGIC SQUARES AND COMBINATIONS.

the remaining numbers

181

may

be

filled in at

random

as explained in

connection with the

8X8

Indian square (Fig. 283).

Two

other examples of the Frierson squares showing inter-

esting constructive features are given in Figs. 289 and 290.

The

/Z7 /03

9Z
/OS

?2 /S

/zs /o?
22,

/Z2
2-4

S30
7>

77
/6
//J

33

<so

/zo

4-$

"7
/o

S'2.

//

/oo //

Fig. 293.

scheme followed

in these squares

may

also be

employed

in

making

magic rectangles, two examples of which are given in Figs. 291 and 292. In Fig. 292 the numbers are arranged in the following
series before they are entered in the rectangle:

l82

SOME CURIOUS MAGIC SQUARES AND COMBINATIONS.

1
.

9
IO
ii

13

17

21

25

29 30
31

2
3

6
7

14
15

l8 19

22 23

26

12

16

20

24

27 28

32

Figs. 293 and 294 are ingenious combinations of 4

also devised

X4

squares

by Mr. Frierson.
features.

Fig. 293
It

is

a magic cross which

possesses

many unique

contains the almost incredible

number of 160,144

different

columns of twenty-one numbers which

sum

1471.

Fig. 294.

Some

of the properties found in the magic pentagram Fig. 294

may
162.

be stated as follows:

Each
It

4X4
324,

rhombus

is

perfectly magic, with

summations of

therefore follows that from any point to the next the

num-

bers
is

sum

and also that every bent row of eight numbers which

parallel with the rows from point to point sums 324.

In each

4X4 4X4

rhombus there are

five others

of 2

whose

numbers sum

162, also four others of 3

3,

the corner numbers of

which sum 162.

In each

numbers,

viz.,

rhombus, every number ends with one of two o and I, 2 and 9, 3 and 8, 4 and 7, 5 and 6.

SOME CURIOUS MAGIC SQUARES AND COMBINATIONS.

Modifications of the concentric magic squares (described in the
first

chapter) have been devised by Mr. Frierson, two examples of

in Figs.

which are shown

295 and 296.

//

184

SOME CURIOUS MAGIC SQUARES AND COMBINATIONS.

eighteen constructive diagonals possess the continuous feature of

the Indian squares, as

^~

shown by dotted
trip

lines, a-nd it-k> possible that

in thr a

by proper trnnnpnnitirm of
tho diagonalo
i

numhm

^ subdivisions

be nude conliuuoua throughout the square:

j**

-^ /y'^^f.^at^f^ **
'2^*^X-0sC^^'

'^&LSi^&^Sf

>

:r^-/>
<
("}

*f<t

CHAPTER

VIM

NOTES ON VARIOUS CONSTRUCTIVE PLANS BY WHICH MAGIC SQUARES MAY

BE CLASSIFIED.

AN
-**

the square

odd magic square must necessarily have a central cell, and \i is to be f*iiWt,1:his cell must be occupied by the
series,

middle number of the

around which the other num(

bers must be arranged and balanced in pairs, the

sum

of each pair

being

n--\-\.

Although

in 5

and larger odd squares the pairs

of numbers are capable of arrangement in a multitude of different

ways
is

relative to each other as pairs, yet

when one number

two numbers

of a pair

located, the position of the other

number becomes

fixed in order
that are

to satisfy the rule that the

metrically equidistant

symfrom the center number must equal twice that

sum

of any

number, or

/r-f-1.
is

In an even magic square, however, there

no central

cell

and
is

no middle number

in the series, so the

method of construction

not thus limited, but the pairs of numbers which

sum

n~-\-\

may

be harmoniously balancqd either around the center of the square,

as in

odd squares, or

in a variety of other

ways.

Mr. L. S Frierson has cleverly

types of even squares

utilized this feature as the basis

for a series of constructive plans, according to which the various

may

be

classified.

He
may
~

has shown

eleven different plansLwhich

with

4X4

squares, nnrl

Tr-

be used in connection

These constructive
plans,
clearly

differential^ the

various types of e*pR

squares,

there being for exa

l86

NOTES ON VARIOUS CONSTRUCTIVE PLANS.

for

another plan

the

Franklin squares, another for the

Indiinr

or continuous squares and so forth, so that a knowledge of these

plans

makes

it

easy to classify

all

X4

squares.

Six of the eleven

plans given by Mr. Frierson cover distinct methods of arrangement,

the remaining

&** plans being made up of various combinations.

PLAN NO.
In this plan, which
is

I.

probably the simplest of

all,

the pairs of

numbers

that

sum n +l
2

are arranged symmetrically in adjacent

/6

NOTES ON VARIOUS CONSTRUCTIVE PLANS.

i8 7

According to this plan the pairs of numbers are arranged symmetrically on each side of the central axis, one-half of the elements
being adjacent to each other, and the other half constructively adjacent as shown in Figs. 303 and 304. This arrangement furnishes
the Franklin squares

when expanded

to 8

X 8,

providing that the

numbers

in all

2X2

sub-squares are arranged to

sum 130 (See

Figs. 305 and 306).

If this condition is not fulfilled, only half of

JZ

i88

NOTES ON VARIOUS CONSTRUCTIVE PLANS.

PLAN NO.

5.

The

pairs of

numbers

in this
it

plan are arranged in alternate

cells in the diagonal

columns, and

produces the continuous squares

which have been termed Jaina, Nasik and IndiaH squares. Fig. 309 is the Jaina square as modified by Dr. Carus (Fig. 222, p. 127) and Fig. 310 shows the arrangement of the pairs of numbers. The

Fig. 309.

Fig. 310.

diagram of the Jikliaii square Fig. 282 is a simple expansion of Fig. 310, and the diagram of the Frierson square Fig. 287 shows
a design like Fig. 310 repeated in each of
its

four quarters.

PLAN NO.

6.

Under
rically

numbers are balanced symmet- (\ u around the center of the square, and a^-this arrangement is
this plan the pairs of

common

to

all

fmi full

ulid

squares, tho even

jquuwu oho wing

rr

J*

Fig. 311.

Fig. 312.

Fig. 311 shows

form

of 4

X 4 square, the diagrammatic plan being given

inJPig. 312.

NOTES ON VARIOUS CONSTRUCTIVE PLANS.

189

PLAN NO.
Magic squares on
this plan

7.

are formed by combining plans

Fig. 313-

Fig. 3I4-

Xos.
314.

to 3, a square

and

its

diagram being shown

in

Figs 313 and

This plan covers another combination^ of plans

Figs. 315

and

3,

and

and 316 show square and diagram.

Fig. 315-

Fig. 316.

PLAN NO.
This
is

9.

a combination of plans 2 and

in Figs.

3,

a square and

its dia-

gram being given

317 and 318.

Fig. 317-

Fig. 318:

190

NOTES ON VARIOUS CONSTRUCTIVE PLANS.

PLAN NO.
This
is

IO.

also a combination of plans 2

and 3 and

is

illustrated in

Figs. 319 and 320.

/J

Fig. 319.

Fig. 320.

is

square

is

made

in accord-

ance with plan No.

2,

but in the

tewr

half the pairs of

is

numbers
any plan

are located apart by knight's moves, which

hitherto considered.
Tt
ir

different to

imin

Fig. 321.

Fig. 322.

Figs. 321

and 322 show

even squares larger than 4

great diversity of design.

A
f

fun ciiamploo or 6

4 these plans naturally exhibit X 6 squares with

Their respective plans are

clusive, in tt
i

given^m

Figs. 323' -3^, to

\jf

e^c^C^*^ <&pj

if

3^, %g\. inct~> &a^3y/

C<J-**^ e"'

NOTES ON VARIOUS CONSTRUCTIVE PLANS.

Figs. 337

and 53$ are

identical with

6X6

squares shown on

pages 19 and 24.

All squares of this class have the

same charac-

190

NOTES ON VARIOUS CONSTRUCTIVE PLANS.

PLAN NO.
This
is

10.

also a combination of plans 2

and 3 and

is

illustrated in

Figs. 319 and 320.

/2

NOTES ON VARIOUS CONSTRUCTIVE PLANS.

Figs. 337

and 33^ are

24.

identical with

6X6

squares shown on

pages 19 and

All squares of this class have the same charac-

192

NOTES ON VARIOUS CONSTRUCTIVE PLANS.

in

worthy of notice apart by knight's moves.

having eight pairs of numbers located Figs. 53^ ^&, and igg illustrate another
It will

6X6

square with

its

plan and numerical diagram.

be seen

Z8
23

33

6
8

J/ SO

/6

it IS 19 $0

NOTES ON VARIOUS CONSTRUCTIVE PLANS.

CHAPTER

IX

THE MATHEMATICAL VALUE OF MAGIC

SQUARES.

THE following from paper

Vol.

quotations bearing on the above subject are cop-

entitled "Magic Squares and Other Problems a ied on a Chessboard" by Major P. A. MacMahon, R. A., D. Sc., F.R.S.,

published in Proceedings of the Royal Institution of Great Britain,

XVII, No. 96, pp. 50-61, Feb. 4, 1892. "The construction of magic squares is an amusement of great
;

antiquity

we hear

of their being constructed in India and China

before the Christian era, while they appear to have been introduced
into

Europe by Moscopulus who flourished

at Constantinople early

in the fifteenth century."

"However, what was at first merely a practice of magicians and talisman makers has now for a long time become a serious Not that they have imagined that it study for mathematicians.
would lead them
to anything of solid advantage, but because the
it

theory was seen to be fraught with difficulty, and

possible that

was considered

some new properties of numbers might be discovered

This has
in

which mathematicians could turn to account.

proved to be the case, for
has been found to

fact

from a certain point of view the subject be algebraical rather than arithmetical and to be
and the 'theory

intimately connected with great departments of science such as the

'infinitesimal calculus,' the 'calculus of operations/

of groups.'

'

"No

person living knows in

how many ways

it

is

possible to

fact is form a magic square of any order exceeding 4X4that before we can attempt to enumerate magic squares we must

The

see our

way

to solve

problems of a far more simple character."

THE MATHEMATICAL VALUE OF MAGIC SQUARES.

"To

195

say and to establish that problems of the general nature

of the magic square are intimately connected with the infinitesimal calculus and the calculus of finite differences is to sum the

matter up."

It is

therefore evident that this field of study

if

is

by no means

limited,

and

this

may

be said in connection with magic squares

the statement will naturally apply with a larger

meaning

to the

consideration of magic cubes.

INDEX.
Abraham ben
Acoustic
163.

Ezra, 124.
Chladni,
117-120,

Compound magic

squares, 44ff.
47ff.
;

figures,

Concentric magic squares,

ifications of, 183.

Mod-

Alternation, Squares constructed by,

I02-II2.

Confucius, 123. Constructive plans, Notes on various, i84ff.

Andrews, W. S., Archimedes, 166.

Archytus,
158.

113, 114,

125,

126.

Grantor, 164.

Aristotle, 161, 163, 166.

De
10.

Arrow heads

indicating sequence,

Astronomer Poet (The) Omar,

167.

la Hire's method of constructing odd magic squares, 13-14, 17; even magic squares, 34ff.

Donaldson, i6in. Diirer's picture, Melancholy,

Babylonia, Magic square in, 123. Bachet de Mezeriac's method of constructing odd magic squares, 17-18. Batteux's series of the Pythagorean

156, 157.

Egypt, Magic diagrams Even magic cubes, 76ff.

ff.
;

in,
;

156.

squares by

De

squares, 18 la Hire's method,

harmonic
106,

scale, 164-165.

34ff.

Binate transposition, Alternation by,

Ezra,

Abraham

ben, 124.

in.
;

in odd magic squares, 7 Examples of, 8. Browne, C. A., 168, 169 Magic square (27X27) of, 160, 168.
;

Breakmoves

Falkener, Geo.

Games Ancient and

(o, ro,
i,

Modern, 174. Figures, Order of

ri}, 79,

Burge,

164,

165.

Franklin, Benjamin, 89, 93, 94, 105, 112; Letters and Papers on Philovi, i, 79, 83, 87, 112,

Caissa, 176; Squares of, 175.

Carus, Dr. Paul,

sophical Subjects by, 89; Parton's Life and Times of, 96.

128, 163, 172, 176.

Franklin

Squares, 88,
179,
184,

89ff.,

94,

95,

Chessboard, Magic squares and other

in,

177,

185;

Properties

problems on

a, 193.

of, 96, 98.

China, Magic square in, 122, 125, 193. Chinese Philosophy (Dr. Paul Carus)
in.

Frenicle, 89. Frierson, L. S., 62, 155, 173, 177, 179, 180, 183, 184 ; Magic cross of, 181
;

Chinese Reader's Manual (Mayers),

123.

Pentagram

of,

182;

Squares

of,

Chladni acoustic figures, 117-120,

Cicero, 161.

163.

177-188. Frost, Rev. A. H., 175.

Fuh-Hi,

122.

Collinson, Peter, 89.

Fujisawa, Prof., 124.

198

INDEX.
Magic rectangles, 181. Magic squares, and combinations, 173 ff. and other problems on the chessboard, 193 and Pythagorean
; ;

Games Ancient and Modern (Geo.

Falkener), 174. Golden Verses, 159. Great Britain, Proceedings of the

Royal Institute

of, 193.

numbers,
I02ff.
;

56ff.,

175; by alternation,
44ff.
;

Gwalior, India, 176.

Compound,
;

Concen;

tric, 47ff.

Harmonic Harmonic
164.

figures, 157.
scale,

Concentric modified, 183 Construction of, 14, 54ff., 185 ff.

I
;

Pythagorean,

163,

Definition of,
i;

Earliest record of,

34ff.;

Even,

i8ff.,

Franklin, 88,

Harmony of the spheres, vi. Hire's (De la) method of constructing even magic squares, 34^.; odd magic squares, 13-14, 17Ho, Map of, 122.

8gfi., 94, 95,

in; Franklin analyzed

;

by Dr. Carus,

Frierson, 177; 96ff. Frierson's analysis of, 129; Indian

(La Loubere), 176;

I25ff.,

Jaina, 87, 94,

176,

177;

in

symbols,
of,
;

120;
ff.
;

Mathematical

study
in, 4,

129

Number
India, 123, 125, 176, 193.

series in, I48ff.

Odd,

iff.

Indian magic squares, 176, 177, I79>

185, 187, 188.

Knight's
tions

move

5-7;

Varia-

Jaina, inscription in Khajuraho, India, 124; square, 87, 94, I25ff., 176,

177

square modified by Dr. Carus,

127, 177, 188.

in, 140; with predetermined summations, 54. Map of Ho, The, 122. Mathematical study of magic squares, value of magic squares, 193. i29ff. Mayers, 123. Melancholy, Picture of, by Albert
;

Jones, Sir William, 175.

Diirer, 156, 157.

Jowett, 158, 161.

Mezeriac's

(Bachet de) method of constructing odd Magic squares, 17.

193.

Kensington Museum, South, Kesson, Mr., 176.

175.

Moscopulus,
don),
175.

Museum, South Kensington, (Lon-

Khajuraho (India), Jaina

in 125.

inscription

Kielhorn, Prof., 124, 125.

Nasik squares and cubes,

175.

Knight's
4, 5-7-

move

in

magic squares, The

Number

series,

I48ff.

Odd magic
La Loubere,
..Subjects
89.

cubes, 64ff.

General notes

176.

on, 84ff.

Letters and Papers on Philosophical

Odd magic
17;

squares,

iff.;

Bachet de

(By Benjamin Franklin),

Mezeriac's method of constructing,

Breakmoves
in,
7.

in,

7;
;

Examples

Life and
lin

Times of Benjamin Frank(By James Parton), 96.

122.
v,

of breakmoves
ciples of,

General prinpoet, 167.

i,

Logan, Mr., 89, 91. Loh, The Scroll of, Lusus Numerorum,

Omar, the astronomer

Orders of
figures,
168, 171.

(o, ro,

ri,}

79,

MacMahon, Major P. A., 193. Magic cubes, Characteristics of, 64;

General notes on, 84ff.
;

Parton, James, 96, 100.

Even, 76ff

Pentagram, Magic, 182. "Phaedrus" of Plato, 158.

Philolaus, 158, 167.

Odd,

64ff.

INDEX.
Philosophical Subjects by Benjamin Franklin, Letters and Papers on, 89.
Schilling, Prof., 124. Schleiermacher, i6in.

199

Philosophy, Chinese (Dr. Paul Carus)

in.
;

Pythagorean,
;

158.

Plato, I58ff. 169. Platonic school, 167.

Schneider, 16111. Schubert, Prof. Hermann, 44, i6in. Scroll of Loh, The, 122.

Plutarch, 156, 159, 160, 164, i66n. "Politics" of Aristotle, 103.

Smith, Prof. David Eugene, 124. Soul, Procreation of the, 159, 164,
166.

Predetermined summations, Magic

squares with, 54. Proceedings of the

"Soul

of

the

World and Nature"

Royal Institute

of Great Britain, 193. "Procreation of the Soul" (Plutarch),

159, 164, 16611.

(Timaeus), 164. South Kensington don) 175.

Spheres,

Museum
of the,
vi.

(Lon-

Harmony

Stifelius, 92.

Pythagoras,

v,

123,

124,

157,

166;

Summations, Magic squares with

predetermined, 54. Symbols, Magic squares
Tetractys, 159, 161. "Timaeus" of Plato,
i66n.
in,

Harmonic
of,

scale of, 163-164; School

of,

Philosophy Pythagorean numbers,

157;

158.

120.

i56ff.

Quaternate transposition, Alternation

by, 109.

158,

159,

164,

Queen, The,

175.

Timaeus the Locrian,

Transposition,
112.

164.

Alternation

by,

106-

Rectangles, Magic, 181. Reflections on magic squares, 79, 87,

H3ff.

Various constructive plans, Notes on,

of
Plato,
158, 163, 166,

"Republic"
167, 168.

Ventres, D. B., 86.

of

Royal Proceedings
Scheffler's

Institute

Great Britain,
193.

Verses, Golden, 159.

Virgil, 124.

of,

(Prof.)

method of con14.

Yang and
Yih King,

yin, 122.
122.

structing magic squares,

Open Court Mathematical

ESSAYS
ruff

Series

ON THE THEORY OF NUMBERS. I. CONTINUITY AND IRRATIONAL NUMBERS. II. THE NATURE AND MEANING OF NUMBERS, by Richard Dedekind. Authorized translation by Wooster WoodBeman,
Pp. 115.
Professor of Mathematics in the University of Michigan. Cloth 75c net. (35. 6d. net.)

1901.

ELEMENTARY ILLUSTRATIONS OF THE DIFFERENTIAL AND INTEGRAL CALCULUS, by Augustus DeMorgan. New edition. 1899. Pp. viii.,
144.

Cloth $

i.

oo

net.

(43. 6d. net.)

ON THE STUDY AND

DeMorgan.
(45. 6d. net.)

DIFFICULTIES OF MATHEMATICS,
edition.

Second reprint

1902.

Pp.

viii.,

288.

by Augustus Cloth $1.25 net.

A BRIEF HISTORY OF MATHEMATICS.

Professor

An

authorized translation of Dr.

Karl Fink's Geschichte Der Elementar-Mathematik, by Wooster Woodruff Beman, Professor of Mathematics in the University of Michigan, and David Eugene Smith,

of Mathematics in Teachers'

College,

Columbia University.

Second

revised edition.

1903.

Pp.

xii.,

343.

Cloth $1.50 net.

(55. 6d. net.)

THE FOUNDATIONS OF GEOMETRY,

fessor

by David Hilbert, Ph. D., Proof Mathematics, University of Gottingen. Authorized translation by Prof. E. J. Cloth $1.00 Townsend, Ph. D., University of Illinois. 1902. Pp. vii., 143.
(43. 6d. net.)

net.

LECTURES ON ELEMENTARY MATHEMATICS,

From
'

the French

by Thomas

J.

McCormack.

Second

edition.

by Joseph Louis Lagrange. 1901. Pp. 172.

Cloth $1.00 net.

(45. 6d. net.)

EUCLID'S PARALLEL POSTULATE: ITS NATURE, VALIDITY PLACE IN GEOMETRICAL SYSTEMS. Thesis presented to the
,

AND
Philo-

sophical Faculty of Yale University for the Degree of Doctor of Philosophy, by John William Withers, Ph. D. Principal of the Yeatman High School, St. Louis, Mo. Cloth, net $1.25. 1905. Pp. vii., 192. (43. 6d. net.)

MATHEMATICAL
man, by Thomas Cloth 750 net.
J.

ESSAYS

Professor of Mathematics in the

by Hermann Schubert, Johanneum, Hamburg, Germany. From the GerMcCormack. Second edition. 1903. 37 cuts. Pp.' 149.

AND RECREATIONS,

(33. 6d. net.)

SPACE AND

GEOMETRY IN THE LIGHT OF PHYSIOLOGICAL, PSYCHOLOGICAL AND PHYSICAL INQUIRY, by Dr. Ernst Mach, Emeritus
-

From the German by Thomas J. McCorProfessor in the University of Vienna. Cloth, gilt mack, Principal of the La Salle-Peru Township High School. 1906. Price, $i.oonet. Pages 143. top. (S s net.) GEOMETRIC EXERCISES IN PAPER FOLDING, by T. Sundara Row. Edited and revised by Wooster Woodruff Beman, Professor of Mathematics in the University of Michigan, and David Eugene Smith, Professor of Mathematics in With 87 illustrations. Teachers' College of Columbia University. 1901. Pp.
xiv.,

148.

Cloth $1.00 net.

(43. 6d. net.)

A SCRAPBOOK OF ELEMENTARY MATHEMATICS, NOTES, RECREATIONS, ESSAYS, by William F. White, Ph. D., State Normal School, New 12 mo. Paltz, New York. Frontispiece and 70 diagrams and other illustrations.
248 pp.,
cloth binding, gilt top, price

$1.00

net.

(55 net.)

THE OPEN COURT PUBLISHING

378-388 Wabash Avenue

CO.
CHICAGO,
ILL.

PORTRAITS OF

EMINENT MATHEMATICIANS
Three portfolios edited by DAVID EUGENE SMITH, Ph. D., Professor o Mathematics in Teachers' College, Columbia University, New York City. In response to a widespread demand from those interested in mathematics and the history of education, Professor Smith has edited three portfolios of the portraits of some of the most eminent of the world's contributors to the mathematical sciences. Accompanying each portrait is a brief biographical sketch,
with occasional notes of interest concerning the artists represented. The pictures are of a size that allows for framing (11x14), it being the hope that a new interest in mathematics may be aroused through the decoration of classrooms by the portraits of those who helped to create the science.

Twelve great mathematicians down to 1700 A.D.: Thales, Pythagoras, Euclid, Archimedes, Leonardo of Pisa, Cardan, Vieta, Napier, Descartes, Fermat, Newton, Leibniz.

*^QK

**

OLIO

INO.

PORTFOLIO
infinitesimal

most eminent founders and promoters of the Johann and Jakob Bernoulli, Pascal, L'Hopital, Barrow, Laplace, Lagrange, Euler Gauss, Monge and Niccolo
calculus
:

No. 2. The

Cavallieri,

Tartaglia.

PORTFOLIO

No. 3.

portfolios especially portraits of

Eight portraits selected from the two former, adapted for high schools and academies, including

THALES with whom began the study of scientific geometry; PYTHAGORAS who proved the proposition of the square on the
EUCLID

hypotenuse;

whose Elements of Geometry form the basis of all modern text books; ARCHIMEDES whose treatment of the circle, cone, cylinder and sphere influences our work to-day;

DESCARTES
schools
;

to

whom we

are indebted for the graphic algebra in our high

NEWTON

who

generalized the binomial theorem and invented the calculus-

NAPIER who invented logarithms and contributed to trigonometry; PASCAL who discovered the "Mystic Hexagram " at the age of sixteen.

PRICES
Portfolio
i

Portfolio i Portfolio 3 (8 portraits) Portfolio 3 (8 portraits)

or 2 (12 portraits) on Japanese vellum, size 11x14, $5.00; both for $8.50. or 2 (12 portraits) on American plate paper, size 11x14, $3.00; both for $5.00.

on Japanese vellum, size 11x14, $3-5<>; single portrait*, 50 cents. on American plate paper, size 11x14, $2.00; single portraits, 35 c.
is

"

The

issue of this fine collection

equally creditable to the expert knowl-

edge and discriminating taste of the editor, Professor David Eugene Smith, and to the liberality and artistic resources of The Open Court Publishing Co." F. N. Cole, Editor American Mathematical Bulletin, New York.
" The selection is well made, the reproduction is handsomely executed, and the brief account which accompanies each portrait is of interest. Prof. Smith a valuable service to all who have interest in mathematics, by has rendered editing this collection. Wherever mathematics is taught, these portraits should adorn the walls." William F. Osgood, Cambridge Mass.

The Open Court

P.

Publishing Co.
CHICAGO,
ILL.

O. Drawer F.

378-388 Wabash Avenue

UNIVERSITY OF CALIFORNIA LIBRARY

BERKELEY
Return
This book
is

to desk

from which borrowed.

the last date stamped below.

NCI MFFRING
NOV
1

DUE on

Lf

15

MAR 2 1 1952

LD

21-100m-9,'48(B399sl6)476
'

YC 32558

868516

THE UNIVERSITY OF CALIFORNIA LIBRARY