MnrcA 17, T9eS.

JOURNAL OF THE SOCIETY OF ARTS.

. Tbe Chairman, in introducin; the reader of the

papm said he was known to many present as baring
served with Krcat distinction as lmctcriojogist and
chemical analyst to the Government of tbe Central
}'sorinccs iu India ; he. was particularly selected for
the ?.j'roimment do a&ount of bis j'reat distinction
at Cambridge in thC study of natural science,
:'.nd in the performance of bis highly responsible
dutiu in Tndin be bad j'ireu the most complete
.satisfaction. His headquarters were at Agm, where
' he had al: sc.rts of temptations to study tbe various
artistic beauties of India. He bad particularly
devoted bimseli to the subject d the paper, and the
discovery whict he bad to describe was one of very
great interest, botb practical aod theoreticaL
.

'
'

The papfr read was

OX SOME DISCOVERIES OF THE

METHODS OF DESIG3' EMPLOYED
IN 3IOF[AMME1JAX" ART.
By E. H. HAXKW", .M.A.
(Late Fcijow of St. John's Coljggc, Cambridge, Fellow of
.
Allahabad Unhersitr.)

The most striking peculiarity Of Mohatnn)cdan or Saracenic art is the employment of

extremely complicated geometrical patterns.
The acwal methods by which these patterns
.

wc:c drawri aiid designed, are, so far as I am

mmre, unknown to modern artists, either in
Europe, in India, or in Egypt, The works of
Prisse JJ'Avesnes, of Bourgoin, or of Gayct,
dealing with Arabian art, are equally destitute
of ariy satisfactory explanation of the matter.
.

'

Some years
ago I conmenced
to study
the
subject
in h:dia.
At kngth, by
a lucky
chance, I discovered a clue in a small room
in on? of the palaces of Akbar, the great
Mogul Emperor. Here, nearly hidden by
dust and dirt, I found the actual construction
lines used by the artist some four and a-half
centuries ago in producing an arabesque
pattern. By meao3 of the clue thus obtained,
it became easy to draw the more complicdted
of the patterns. The clues to the simpler
classes of pattems were obtained for the most
part by cbseNation and medsurement.
The methods that I have to describe to-r.i;,"ht,
are not simply n)ethods of draughtsmant.hip.
They may be described as methods of design,
as by their means it is possible and eacy, not
only to copy old patterns, but also to design
new ones in conformity with the rules of
geometrical art.
It may be advisable to explaio that the
patterns themselves, as I shall exhibit them to
you th.ro:':n gn .the s,creeo as black lines Qr0 a

q6x

white ground, are not necessarily or essentially

beautiful. They are merely skeletons of
beautiful objects. An artist's skill is required
to clothe wi:.h' flesh, the dry bones of geometric desigq to produce from them artistic
creations that impress the feelings and that
exist as living triumphs of Saracenic art in the
frescoes arid tilcj dadoes of the Alhambra, in
the mosaics, pulpits, and rnosques, of Cairo,
and in the deserted palaces of Akbar at
Futteypur-Sikri.
.
Geometrical patterns may "be divided into
the following four classes(i) squ:zre, (2)
hexagooal, (3) octagonal, and (4) arabesques.
0

Fig. i.

)))))1)))))

).))-))j,|)-))))))')j'-)
Hexagonal A'^"n"ERN froji Hakim's Bath at
FUTTEYPUR-SIKRI.

In the first class the space to be decorated '

is divided into squares. Parts d these squares
go to form the pattern.
This class wiii be
found described in any elementary text-book of
design, and includes various chequers, the
fyifots,fhe Greek frett, rectangular lattices, &c.
The method of drawing the second Class of .
patterns, the hexagonal, is also widely known
,
and calls for no special deScription. Lines are
drawn crossing each other, not at right angles,
. as in the square patterns, but at angles of
sixty degrees, thus dividing the space to be
'
decorated into equilateral triangles. An example of this pattern is given in Fig. i, which
represents a design found carved in plaster in
the Doctor's Turkish Bath at Futteypur-Sikri.
In this case some only of the Qattcrn lines are
identical with the construction lines. Others
' of the lines of the pattern are drawn crossing

JOURNAL OF THE soc1E7i" OF ARIA'.

462

the construction triangl.es from one apex to

aoother.
The third class of patterns consists of, or is
derived from, octagons. Simple examples of
this class are shown in Fig 2, at A,]3, C, and
d. Of these a, b, and C may be found used
in stoae pavements in the garden of the Taj at
Agra. d is frequently employed for the per[orated stone balustrades on buildings in the
Taj Garden. I have grounds forbeliei'ing that
in drawing octagonal patterns an octagon was

Mara 17, x9q$

Indeed it is diflicult to see how the more complicated of the octagonal patterns could have
been desigoed without the use of templates.
The complicated octagonal patterns usually
contain octagons of two sizes, which, so far as
my experience goes, may always be regarded
as derived one from the other by the following
very simple construction. In Fig. 3 an octagon
is drawn in which opposite aogles are sbown
joined in pairsbylines such as A R,and C D. The
result ofdrawing these linesis to leaveao eight-

Fig. 2. .

) ")

))/
i

Examples OF OctaGonal Patterns.

CUt out of talc, or paper, or some other suitable pointed star (shown by thicker lines) in the
material, and used as a template.
In the
centre of the octagon. Taking the point A as
case of Fig. 2, D, the octagons shown do the centre describe another octagon (shown l?y
right of the figure are so accurately drawn (as . dotted lines) of such a size that one of its angles
a rule) as to suggest the use of a template,
fits into the angle E of the eight-pointed star.
while the remaining lines are so inaccurately This octagon is shown drawn separately at G.
drawn as to suggest that they were filled.in by Thus we have a large octagon, a small octagon,
hand. Whether or not a draughtsman should
and an eight-pointed star. These three fij'ures
use templates in the actual employmen: of combined in dierent ways go tO form the
these patterns may be deScribed as a matter majority of octagonal patterns. For drawing
of individual taste and habit. But that temthe patterns it will be found convenient to cut
plates are useful in the designing of these
out these three ytiines as templates.
patterns xviii be admictcd on consideration of . The method of construction of a patt.em of
the facts contained in the fdiowing paragraphs.
this nature is showa in Fig. 4. The large

.. Jug AiAb4'AA

.|;,791.^ S;, !y'3.

P

.'

A iim j. - ". '

4'"j

"2

aTe drawn whose centres coincide with alternate

angles of the large octagon, as shown. on the
left-hand side of the figure. But this does not
produce 'a graceful pattern. The numerous
lines crossing at a point,'as occurs at k and ,
.at l, like the spokes of a cart-wheel,. do not
form a pleasing feature. >"or is the case much
altered for the better when a square is drawn
in the small octagon, as shown at m and N.
But when cert3in of the lines are omitted, as
illustrated in Fig. 5, we have a pattero of remarkable gracefulness io spi.te of its appareat
simplicity. This pattem occurs carved io low
relief Id a panel of red sandstone in Birbul's
House, F.itteypur-Sikri.

octagon
template
is first
employed
to cover
the ground
with large
octagons
overlappioy
by
two of their sides, as shown on the right-hand
side of the figure. For the sake of clearness
the letters a, b, c, d, e, f, g and h, have
been inserted to indicate the extent of one of
these octagons. The small octagon template
must now be used. By its meaos small octagons

Fig. 3.
.

""""ii"
Fig. C.
.

6?

'

'E ?:

Df.riv4t(on' OF Small Octagon FROM Large

Oc]7.g')n
also formj
of Eight-pointed
Star.
These three ;ligures
the basis of most
oct-

si

?:tay,
:!!r\ i>Q

agonal pattems.

Fig. 4. '
~. ,

.
/

.
L

,L7~rcs

Pattern from a Pilaster in Blrbul's HOUSE,

'
FUTTEYPUR-SIKKL
.
It may 'be asserted that in this pattern the
size of the octagon is well proportioned to the

=0"A"(
\7jym7mr
"I

siz.of thepane1,andthesi,eofthes,ua,e

ii

\c /

. V4'L^ 1'~

lg

that ofthat
thethis
octagon.
It is difficult
to .
,. .,'O
imagine
result would
have been
obtained so successfully had the design been

\F/ I
.

, ) E

"

is shown "in Fig. 6.

In this case the large
octagons are drawn, not in contact, but at
Fomc distance from each other. The distance

"R'
.

"

/ \

formed without the help of a rational geometrical method.

The construction of another octagonal pattern

/\~~U\~

adjacent
octagons
could
just be any
filled
by
is such that
the space
between
four

Construction of Pattern shown in Fig. 5,

The ground is covered with large octagons overlapping by two side3. Tbe letters A to E itidicate
Lac cutiine of one of these oct3gons.
Small
octagoos, as at K, L, M, N, are drawn round

another ]arj"e octagon. The small octa gon

template is now brought into use. As shown
on the right-hand side of the figure, it is placed
with its centrt' coinciding with the angle of a
large octagon. Six sides of the small octagon

are drawn iii.

allemate angles of the large octagons.

The operation is repeated round

JOURNAL OF THE SOCIETY OF ARTS. '

464

basis of a pattern that occurs ia a dado in the'

vestibule of Akbar's tomb at Sekundra. Jn
this case some of the eight-pointed stars are
drawn complete. Others are overlapped by
their neigbbours.
Pattern g is obviously
formed frcm eight-pointed stars. A very beauttfu] example of the use of this pattern will be
found io the work if Owen Jones on the

Fig. 6.

XC

\\'e now have to consider the most complicated class of patterns to which the term
" arabesque " may be appropriately reserved,
as these patterns are practically peculiar to

g;q3A(
Coxstkcctjox" OF Octagonal P^tter SHOWN
co3iplete in FIG. 7.

As shown at A, six sides

of a small octagon are drawn round the angle of a

Tbis is rep2ated rcund a!] tbe

other angles (as B, C, D) of the large octagons.

"

Fig. 8.

\ )1='\- 3,|

large octagon.

Alhambra. H is a. common border pattern

" formed from eight-pointed stars.

jj

Mb.rh 17, xgo

0

each of the :othcr afgies of the lar'ge oct3gon.

Thereby the eight-pointed star is found to have
been described in the large octagon. The
other large octagons are similarly treated. A
few other lines are required to complete the
pmem. These are so simply placed as to

,'

Tbe few Hoes required to complete the pattern may""

be found by insction of Fig. i.

(I));,

Fig. 7.

Diagrammatic Sketch OF PART OF Pattern

OF A. Dome IN THE TURKISK BATH IN THE
" Jodh Bai's Palace AT FUTTE\"PUK-SJKRI.
Tbe Hoes A, B, C zre some of the original codstruction lines by means of wbich the paltem was
drawn. Examples of the use of this clue are given
in the following GgUres.

|n":))'))ji:

Octagox"al Pattern OCCURRJXG IN Inlaid

Marjsle IN THE Tomb OF I'I'JMAD-UD-DAULA
AT AgRm

require no explanation. The completed design

is shown in Fig. 7. It occurs in inlaid marble
in the tomb of Itimad Ud Dada at Agra.
The elm-pointed star alone, without the help
of any octagons, may also be used to form patterns.
Examples of this are shown at e, f, g,
and h in Fig. 2. In e eight-pointed stars are
placed in contact. This is the pattern of the
pavement of the interior of the Taj. f is the

'

'

Mohammedan art, and both in their complexity

and in their :tsthetic effect stand on a higher "
plane than the patteroS hitherto considered.
In most square, hexagonal, .and octagonal
'patterns, the pattern lines run in two, three, or '
four directions respectively. In arabesque
patterns the lines run in a much larger n:imber '
of directions, suggesting that some unusual
method was en)p)oyed in their construction.
In Fig. 8 is represented diagrammatically
part of an arabesque pattern that occurs in a
dome in the Turkish bath attached to the Jcdh
Bai Palace, Futteypur-Sikri. It was this
pattern that furnished the clue jo the method
of construction of arabesque designs. Besides

"Mara x7, 1905

JOURNAL OF THE SOCIETY OF AR7S.

the thick pattern lines incised in the plaster

covering the interior of the dome, I noticed
some faint scratches indicated by the lines a i:,
and B C. On closer examination, I found
that these scratches were parts of polygons,
that if completed as shown at D, E, F, would
surround the five, six, seven, and eight-rayed
stars of which the pattern was composed. It
at ooce occurred to me that these polygons
were the construction lines used in producing

465

A few examples will show how simple is

the method here indicated for drawing and dcsigninj" of arabesque patterns.
In Fig. 9 the space is first covered with
octagons in contact, which may be easily drawn
by means of an octagon template. The resulting lines may be called the primary construction lines. Two pattern lines have to bedran"n
through the centres of each of the primary construction lioes. The first question is at what

Fig. 9.

"G""O"" "" "

>,41> OQ (
( '_J/ "

" j'i "-!>

S.j'~

r)
Method op Construction of a simple Arabesque Patterx". The ground is first cmered with
ocmgons in contact. To form the p.ittcm two lines hare to be drawn through ea:h centre of each
side of the octagons.
.

the pattern. It was obvious that polygons in

contact were easy to draw. Having drawn
them, pairs of lines had to be drawn passing
throngh the centres of each of the sides of the
polygons. These pairs of lines crossed the side
of the polygon, at nearly the same angle in
each case. Each line was"pro1onged until it
met a similar lioe that had crossed the centre
of another side of a polygon. When this had
been done all over the surface that iiaci to be
decorated, nothing more remained to be done,
{or the pattern was completed.

angle or in what direction they are to be drawn ?

Within certain limits the exact angle does
not matter provided this angle is the same in
all cases. But the best result is obtained if
each pattern line is drawn parallel to a diagonal
of an octagon. For instance, the pattern line
a b is drawn parallel to the diagonal C d.
But in practice it will be found easier to draw
lines such as t u which pass through the
ccntre of the octagon and join the ccntres of
its opposite sides. A pattern line is then drawn
sucijasEF. This of course is drawn parallel

466

JOURNAL OF THE SOCIETY OF ARTS.

JL'arrh j7, 3gc$

Cagons in contact. As before, two pattern

lines have to be drawn through the centres of
each of the primary construction lines, and it
is necessary to commence by finding out the
direction in which these lineshave tobe drawn.
C and d are centres of construction lines. A
pattern line a b must be drawn through these
two points. This pattern line is continued
until it meets an interradius of the decagon at

" to the neighbouring diagonal of the octagon.

Itcuts TU in the point E. From the centre Of
the octagon describe a circle g which passes
through the point e. All other pattern lines
pass from points similar to e on the circumference of the circle. A guiding circle inay
similarly be drawn in the squares, as shown at
li.
These circles and interradii form the
secondary construction lines.
Ii] the portion oIl the pattern shown completed, it may be noticed that the pattern line

}3. From the centre of the decagon.a circle h

is described that cuts b. Other pattern lines
meet this circle ,at places where it cuts the
interradii. This .is the most frequent method
of finding the position of the secondary construction lines.

Fig. JO.

Fig. II.

))))):/

U) ,-U'"-'f \J\-

Cox"strcct!ox of Arajcesque based on Decjlgons axd PEXTAGONS in cox'tact.

The
first pattern line is drawn through centres C and D
till it meets a radius of the pentagon at A. - Tbe
second pattern line is drawn from A through the
centre of the side of the pentagon, tin it meets ao
interradius of the decagon at E.
Theze two
pattern lines Rive the size of the circles K and L.
All remaining pattern lines pass in pairs tbrougb
ccntres of sides of a polygon until they reaCh a
0

Arabesque based On Decagons in Contact.

Two lines are dmwn passing tbrough the centre of
each sid? of tb: decagon. Each of these lines is
continued until it meets anotbcr similar line, thus
completing tbe pattern.

lX n is nearly but not quite in a straight line

with the pattern line O p. This line o p is
nearly but not quite parallel to the pattern
line r s. A draughtsman not knowing the
method now described would probably have
djawn ?il x and O p in a straight line with one
another. He would also probably have made
O p and g S exactly parallel. So doing would
have given the pattern an appearance of stiRncss which is avoided by the present method
of construction. This pattern is one of the
' common'est of arabesque designs.
Another arabesque v is shown in Fig. to.
Herc the primary construction lines are dcK

circle. The completed pattern is shou n in Fig. 12.

B

Iii this pattern each decagon of the construction line leads to a ten-pointed star
of the pattern surroynded by pentagons.
' These pentagons are regular if the pattein
has been correctly drawn.
Other similar
ardbesques exist in which the primary construction lines are dodecagons and octagons,
or dodecagons combined with regular ninesided polygons. In these cases the resulting
star-shaped spaces in the pattern are twelve,
eight, and nine P2!"tSd, as the case u!ay be, and
the pentagons dre not'completely symmetrical.
We now pass on to consider an arabesque
' of a slightly higher degree of complexity. Id
'Fig. i! its primary construction lines are
drawn as thick lines, and its secondary con.

March 17, 3905.

JOURNAL OF THE SOCIETY OF ARTS.

structioa lir.es are . drawn thin. The primary

construction lines consist of regular decagons,
regular pentagons, and irregular hexagons.
The templates for this pattern consist of a
decagon and a pentagon.
Or more conveniently a decagon may be drawn with a
pentagon attached to each of eight of its
sides. This may be cut oUt of paper as ooe
template. Such points as are required that
are not on its periphery may be pricked through
on to the drawing paper by means of a fine
needle. The irregular hexagon requires no
template. It is merely the space left where
there is no room for pentagons.
In this pattern the secondary construction
lines are found by the following easy method.
The first pattern line a B is drawn through the
centres of two construction lines at C and d.
It is cootinued till it meets radii of the pentagons at each end,as atAanciB. AcircleL
is described in the pentagon cutting the point
a. A second pattern line a E is now drawn.
This starts from the point A and passes
through the centre of the side of the pentagon.
It is continued until it meets the interradius
f h of the decagon in the point e. A circle
k is tjoiv drawn in the decagon that cuts e.
All the pattern lines have to be drawn .starting
from analogous points on the circles k and l,
or from similar circles drawn in the other
polygons.
The completed patterri is shown in Fig 12.
For the sake of making the pattern easier to
comprehend, alternate spaces are tinted, as
might be the case if it was used in a mosaic
.

centre. The method of construction though

tedious is by no means difhcult. Round the
ccntre a describe a dodecagoo, of which one
quarter is shown at E f h.
Touching this
dodecagon at the point f, another dcidecagon
,of the same size has to be drawn. It is lettered
f K l y and shown drawn round the centre x.
Touching the dodecagon round x at the point
y is drawn a third similar dodecagoo, whose
centre is at O. Touchiog this latter at ? is.
drawn a fourth dodecagon whose centre is at
r. Inside each of these large dodecaE"oos are
to be drawn smaller dodecagons. Their size
is found as follows :join e k. On a radius
.of dodecagon a mark off e b equal to E k.
Fig, Y2.

'1')_)
Arabesque based on Decagons and Pentagons' ix contact. Notice tbe lines A B and
C D, and similarly placed pairs of lines, are not
quite parallel. II they Ind been drawn parallel
the pattern would bare been more stiff and less
jIeasing. Tbe lines E K, L A, F H, M X, P R,
are nearly, but not quite, in the same straight line
with one another. This also prevents stiWness,
and supplies an additional proof Of the admntage
of the method di constructioo explained in this

pavement. It is well to notice that, as here

dr:zwn, the lines ek, la, FH, mn, and pr,
are not quite in a straight line with one
another. Ad artist not knowing the correct
method of drawing the pattern would, with
paper.
little doubt, have made of these lines a hard
straight line running through the design, like Then a b is the radius of the small dodecagon
requited. This is shown drawn and lettered
a line in the pattern of a cane-bottomed chair.
He would have added to this resulting and }3 C d in dodecagon a. A similar small
unnecessary stiffness, by making the rays of dodecagon has to be drawn in each of the
the ten-pointed star of parallel lines, which is other large dodecagons. A remaining feature
equally a mistake, for a ii and C d are nearly . of the primary construction lines is seen as an
irregular octagon at p. This is merely the
but not quite parallel.
spice kfc between the four large dodecagon9
' A still more complicated pattern is shown
A, X, O, and the fellow of n, which wou1d have
in Fig. 13. Though the edect of the pattern,
as here drawn, is surprising rather than been drawn had the adjoining quarter of the
pleasing, in the oripinal, as it exists in a dado panel been represented. Half of this irregular
of coloured tiles in the Alhambra, it is a very ,octagon is shown at e k l m. This sp.ice is
beautiful object. An illustration of it may repeated at p, and elsowhere in the pmcrn.
be found in the work of Owen Jones on the The radii of this octagon must be dr.au'n in .Is
shown at x. No further description is needed
Alhain5ra. My drawing only represents a
quarter of the panel, the point A being the of the primary construction lines.

JOURNdL OF THE SOCIETY OF ARTSJ

.68

pointed star simildr to tMse preseot in other

parts of the design.
At VII and VIII (in Fig. 13) an eight'
poiated star is shown.' The primary construction outline for this star is a square,
shown round VIII. The exact size of the eightpointed star is a matter of tastC. I have dram
it in such a way that the pattern lines I, II,
III-IV, V-VI, are all in a straight line with
one another. The original drawing is not
sufEciently accurate to deterniine ,this point.
In Fig. 14 is represented a panel from the
side of a stone pulpit in Cairo, copied (with
some modification) from Stanley Lane Pool's
work on Saracenic art.

The secondary construction Jiors are very

simple. As may be seen by inspection, a
,

March 17, x9dS.

mimber
of the pattem
lines are
pssing through
the centres
ofdrawn,
two prsimply
more
.construction lines without any further guidance
being required. In dodecagon o is drawn a
twelve-pointed start of normal type.
The
twelve-pointed stars drawn round a, X and R
are modified to give further variety to the
pattern. It May"be well to state that in solving
the method of construction oF this pattern, I
had at my disposal only a very rough tracing
fthe not altogeiher accurate drawing in the
work of Owen Jones. Consequently I had to
rely to some extent on my general knowledge

Fig. t3.

^5='==="'

\ D"" "'""m"""
"'"j

r "1 '

,c"jfr ,J -v

. e,

'<

":'"y\ ':,.... ....<:' / .:4^a a.. . ,.

'b - r'\
.

"

. .

' <' I
0

. ..

..
- Cl

t"

j' :

p
00

00

. .j ..

'0r'

G%L"A^- KA Ml^yArdy7 """"".Y.J

W

'

..

Ei

0 _

':

"'

0 0

< -5-\
'
)l "k.\
>~.

i '. W

..

" i< ' "

jy"u::)'i)ah ")Ti*ry&" mr,

..,..
' ..V

"S '.i I\_L

.
P

LA

~ A~lwi

K.

^. f-""

l.

"\ >

~ i1"

r,

>

Y"mM } L
'"'"' 1
0 "

K ,
a'

, .-....;><1.
~ ."1 ^^l \

/'""X AJ~

^ m k

\-.

0
0

I -""

"i ' t>f ''

.1
>"'""'.
')'
Ll'

L'l

"1~,

.~1 '".r i

"0

\ ' ..'^< m

j1^T\J\/

>- W

0 ,

>6.

'L :~7"'>"""$; P vnX ?

'"jr-'""'"
P

..or '

'

/1A"Y K1"':. ? --"-;""

.....

'

;~-
*.\
"DC " Av-

L
K .

7
' " )'.Y. "Vf

'"7{,.,,':::'.'," '7 X '

\vffV. ... ":?!1,. .. : ,"'|lj:

""m.'."
0

I /f

IL
i ," kn
/"<Y

_ , m "i . .

W4, p

>-Lr"u
0 L.

>?

""""'-O<-:v:'f~

\ Y4-LL:_
0
0

F'0

-'

j........, j^

0
P

0
0

U'

"'

~ T"""'

" """4 1.r"1

' r"
L

-^'

I'

"\ .^-

Ir/''

^: -m ^

Method OF Construction OF Ah" Arabesque BASED Ox Dodecagons, Four-sided Figures AND

Tria.cgles.

From a tiled dado in the Alb3mbra.

One quarter only of the repeat is shown.

it

<ji the subject.

The twelve-poiOted star drawn

For drawing this pattern two templates are

Q

round a has been made of such a size that

the spaces S and t are equal to each other,
,.and also equal to the
Space u.
The

required. One is a regular sixteen-sided polygon.

The other is a regular heptagon, the'
.length of whose side is the same as the length

space "j is made as far as Qossible symmetrical. The twelve-pointed star drawn round
:x is made of such a size, that the space v is

of the side of the sixteen-sided polygon. Per-'

haps for coovenience of description I may be
allowed to refer to the latter as the " i6-gon."

equal to. U. A similar consideration guvemed'

the size of the space x in the star drawn round

With the" i6-gon template draw the outline

B shown Id Fig. iS. Place the heptagon tern-

r.
The irregular octag"on p is adapted to
produce an irregular eight-pointed star. This

platewith one side touching b at k h: Bythis

means draw in the heptagon d e f g h k l.

would be a blemish in the pattern.

Con-

TwO sides of this heptagon, f G and G FI, are

sequently alternate p3ints bare been thrown

back, as shown at x, which has been done in

indicated merely as dotted lines as they are'

not further required.
TJi2 centre of this

such a way as to produce a symmetrical twelve-

heptagon is at n.

A second similarly situated

"

jouKnal OF THE SOCIETY oi' ARTS.

March 37, x9oS.

469

heptagon is now drawn in having its centre at

p. The i6-gon template is now placed touching the sides e f o of the twinned heptagonsb
and a i6-gon is thereby drawn shown at a.
Twinned heptagons are similarly drawn at
symmetrical intervals round the original i6-gon
b, and on one side the twinned heptagons
indicate the position of another i6-gon which
is shown in the 6gure drawn round C.

one ofits sides touches, or is the same as, the

'

side of its fellow.

Tne primary construction lines havC now
been completed. The secondary construciion
lines consist of the radii of the different
polygons, and of one or two circles drawn in
each of the polygons. Within limits the eiact
sizes of these circles do not matter. By lriai
and error a suitable size can easily be found.

Fig. 14.

{X

7"ji'""

)1):)?
i EL4L'
~ .9 "~
"""0,,_
"" "
"~~
:" .=:=
:sm
~==-2m"""
"""^m"'"'j V ~
P

"r

" 3 "'"Zm9'
f

Pattern FROM PAXEL OF A FOURTEENTH CEXTCRY STONE PULPIT IN CAIRO.

(Modified from Stanley Lane Poole.)

0

Between the i6-gons a, b, and C, is left a

space which is filled up by a dodecagon r.
Similar spaces are 61led up by similar dodecagons at s and i,' and also (partly drawn) at
ehe three cgrncrs of the panel. These dodecagons are not quite symmetrical, but must
be drawn as regular as the space available
jmmits.
.
In each of the twinned
heptagons smaller
heptagons have noW to be drawn, such as
othosc shown with centres at x and p. Each
tjf these" heptagons has its centre identical
with the centre of its larger surrounding
beptagon, and is drawn of such a size that
0

In the previ6usly described arabescjues, the

pattern lines were drawn through the centres
of the construction lines. ' In the arabesque
now under consideration this is not the case.
Each primary construction line, that is to say",
each side of a polygon, is to be divided by two
dots into three equal parts. The pattern lines
arc drawn through the dots. Some of the
pattern lines ha,yc been drawn in. as dotted
lines in p.irt ofihe Sgllre. hese will serve as
a guide to enable the student to complete the
pattern with facility.
.
" Supposing it is required to discover the construction lines of any given arabesque, proceed

JOURNAL OF THE SOCIETY OF ARTS.

470

as follows. Jiark the centres of ail the larger

spaces iocluded Id the pattern, excepting, as
? rule, the star-shaped spaces. Join these'
centres. The polygons thus produced are the
primary constructioa lines.
Supposing it is
required to copy an arabesque that exists on a
ceiling, or in some other inaccessible position,'
a similar method will suf6ce. On looking at
the pattern,it is easy to imagioe lines joining the
centres of the larger spaces, which lines describe
.

March. 17, xqo

It will be advisable to conclude with a few

remarks on some of the mies observed in
Mohammedan decorative art.
One of the objects of decoration is to prevent
the eye being dispkased by monotony.-. A
geometrical pattern, however elaborate, if used
too much, will produce that effect of monotooy
that it was intended to prevent. It will then
become as tedious to the eye as a Greek fret
was to Ruskin, or as a cheap Boral wallpaper

Fig. jS.

jtjj\

jjj

2)' '- ' ...,@.:)')?"'.!,3...

j;. M':::.';:!".::"
-"EJ
W I

'E7:-{":"'y'" '

&NA

'"3" ,5a%:

Method of Coxstruct1ox of Pattern shown in Fig. 14. As shown in lower part of the drawing,
each side of a polygon is divided into three equal parts by two dots. The pattera lioes are drawo
through these dots.

polygons.
A rough sketch .may be made of
these imagined polygons. From this, atleisure,
an accerate drawing may be made. In doing
so it is necessary to make the different polygons
as symmetrical as possible, and, so far as
possible, having their sides all of equal length.
Guided by this rule, and after a little practice,
any complicated arabesque pattern' can usually
be solved in ten minutes. On the other hand,
I personally have failed to solve %me of the
apparently more simple' patterns despite a
"more extended study.

may be to anyone else. Jo some of the be'st

buildings in Futteypur-Sikri, the artist has
nearly worked up to the standard of discarding
each pattern as soon as he had used it once.
In Birbul's house, for instance, a number of
richly-decorated pilasters are present in the
different rooms. Each pilaster has, ori its.face
fronting the room,"three carved paoels. Od
each pilaster the upper and lower of the ,three
panels bear the same pattern, but th,e middle
panel always carries a different pattern, and
the patterns used on ,one pilaster, .so far its I

Marrh J7," 1905:

JOUR1\ML OF 777E SOCIET12 OF ARTS.

47'.

dm awdre, d:t not occur on any other. These

along the sides. But the Moorish artist has
patterns are nearly all geometrical, but a few
put more than a quarter of a star in the
are Rora), thus further preventing any imprescorner, and more than half stars along the
sides." The arraogement is such that the
sion of monotony. One of these pilaster patterns
is shown in Fig. 5.
rays of the stars along the sides form a
That the pattern should be adapted to the
sort of, inner border which is accentuated
space it has to occupy is a truism observed in
by the cQouring adopted in the original
:iimost every system of art. In the case of geodesign.
. of this rule a paoel oi
Owing to obsewaoce
metrical patterns, this truism admits of a simple geometrical expression, which, with rare
given shape is not suitable for the reception
exceptions, is adhered to by Mohammedan
of any and every pattern. If the panel is a
artists, and ignored by Europeao artists
square, or if it is made up of a whole number
when copying Mohammedan desigas.
In
of squares, it may be decorated by an octaall the patterus used to illustrate this paper,
gonal pattern, or by those octagonal and
dodecagonal arabesques whose repeat is a
star-shaped spaces occur at regular intervals.
The rule very generally observed .in Moham-. square. The repeat of a hexagonal pattern is
usually a rectangle whoSe diagonal forms an
medan art in India is that each corner of the
panel is occupied by a quarter of one of these
angle of 60 with the base. Such patterns are,
star-shaped spaces. If several repeats of the
therefore, not applicable to square panels, but "
pattern occur in the panel, half stars will occur
only to panels that are rectangles of this
proportion, or to panels that may be regarded
along the sides, besides quarter stars in the
as made up of such rectangles. The repeat of
corners. The panel always contains a wbole
arabesques whose primary construction lines
number of repeats. If the pattern contains
eight-pointed starS, the space included in the
include decagons is usually a rectangle,
stars may be filled up by fylfots (or,outiines of whose diagonal forms an angle of 36.
In
.
that nature), and then adherence to the rule"
" Les Elements .dc L'Art Arabe," by Bowis only to be discerned on studying the con-, goine, may be, found a large collection of
arabesque patterns. In some of these patterns
struction of the pattern. Id some of the perthat contain T4-rayed stars, the unit of pattern
fo:ated stone balustrades that form so pleasing
is
a rectangle whose diagonal forms an anRje
a feat:ire of the buildings in the garden of. the
Taj at Agra, the pattern has been visiblj' and
of about 38. With sufficient knowledge a
,
.
greatly distorted to ensure compliance with . : p.fticrn may be fuund to suit a rectangle gf
this rule in the space available. The tomb of ' almost aoy shape. .
A few words may be added as to the amount
.
Itimyad Lid Dada in Agra oSers a striking.
gf accuracy needed and advisable in drawing .. '
illustration of the observance of this congeometrical patterns.
In elementary textvention. Worked into its marble mosaic
are hexagonal, octagonal, decagonal, and .books of design it is commonly asserted that
pattern outlines look better when drawn in by
dodecagonal patterns, in the greatest prohand than when .drawn .in by compass and
' {usioo, but in every pane], whatever its
rule. The truth of this statement may be
size and shape, quarter stars are always to
admitted without reserve so far as the simplest"
be found in each of its four corners. The diffipattern shapes and outlines are concerned.
culty of designing the building.must have been
.
greatly added to by the observance of this rule. '" With regard to Jess simple patterns, however,
some reservation is necessary. If a hexagonal
One is tempted to suspect that the building
pattern is made with the aid of construction
was designed to 6t the patterns.
,
lines that have not been drawn at exact];}" the
In Mohammedan art as it occurs in Egypt
correct .angle, every hexagon in the pattern
.
and Spain this method of adaptation to- the
will
be
distorted.
If
lopsidedness
occurred
in
' space is sometimes followed. At other times a
a single hexagon it perhaps would not matter.
different plan is adopted, an example of which
But. when every hexagon is lopsided to the
may be found in Fig. 13. As already cxsame extent, and in the same direction, the
plained, this drawing reprcsents only a quarter
effect is cumulative and unpleasant. Thq'
'
of the complete pane]. The border is thereP

fore only shown on two sides, indicated by

-

double lines. Had the pattern been used by

a Mohammedan in India, the comer would
hare been' occupied exactly by a quarter of
a star, and half stars" would have occurred

suggestion that the artist was trying to draw "

hexagons and failed because he didn't know "
how to do so, is a suggestion that otricusly
should be avoided. If, however, the construction lines of a hemgo=U pattern are drawn in
0

,472

JOURNAL OF THE SOCIETY OFARTS.

Mamh x7;.x9oR

faces ' with sculpture, mouldings, and paint, and

with accuraCy, the pattem may wei! bC drawn
mosMcs, encaustic tiles, and inlays of mother of pearl,
in by hand.
'
.
ivory, ebony and metals, and, in the case of carpets,
Arabesques aPe usually found drawn accurhangings, and other textile fabrics [" Turkish tapesately. In the case of the more complicated
try "], with broidery in " blue and purple and
nrabesques, this would seem to be necessary
scarlet."
Tbis is the narrowly limited technical
to produce their full 'esthetic effect. But a
question Mr. Hankin's discorery revives, and yet it is
relatively simple arabesque, such as that
impossible to discus8 it, even within its most closely reshown in Fig. 12, may be distorted in order
stricted technological scope, with any adequate Intelto make it fit the space ; that is to say, to get
Iigence, independently of its relation to the wider
exactly quarter stars in each of the. corners of question of the origins of the inspiration and conthe panel. In order to do this it is necessary
structive features and mechanical methods of " Sarato draw first the construction lines, modified or
cenic " art so-called. The earliest building in Arabh
that can Ge unhesitatiogly attributed to the Shemitic
adapted to the space, and then on these lines
Arabs, as distinguisbed from the Hamitic Adites, is
to draw the pattern.
the Kaaba, or " Cube.house" at Mecca, and as
By observing which patterns were p?aced in
rebuilt
in the days of 3[ahomet himself, it was the
prominent positions aod which were placed
work
of
a Greek, called Bakum by the Arab writers,
where they could not attract much notice, it
and a Copt, They were in charge of a large cargo of
is possible to arrive at some ideas as to what
" Greek jjhss " for the decoration of a Christian .
features of a:pattern were considered good by
church in Abyssinia, and haring been shipWrecked .
' Mohammedan artists. A pattern containing
near Jidda, were impressed by tbe Kboreish Arabs ior
several lines crossing at one point like the
the restoration of the Kaaba. The Caliph Omar, tibe
. spokes of a wheel, as already mentioned, is
successor of Mabomet, built the " Sacred Mosque "
oot much used. The best patterns are those
(masjid'l-haram) round, or rather foursquare, about
in which all the constituent spaces have either
the Kaaba, and Burkhardt was quick to notice that
a radial or a bilateral symmetry, that is to say,
the only good decoration on it was of Greek
are free from lopsidedness. A pattern is bad
design and workmanship, and tbat all that was,
"if it contains spaces of the same shape but
as might be presumed, of natire handicraft, was
most coarsely and ignorantly executed, some ot
differing slightly in size. If a .construction
the pillars of tbe colonnade being set up in their
leads to such a result, the pattern must be
places
upside down.
Second in sanctity to the
modified so that the spaces become either
Mosque
at
Mecca
is
the
" Mosque of the Prophet "
identicalin size or else widely diEerent. Cases
'
(masjid'unnabi)
at
3!edina.
As built by "the

"have been mentioned in this paper in which

erophet of God" himself, it was a plain building D!
'
gracefulness in a pattern is obtained by draw" Sundried bricks, unbewn stone, and date palm stems
.
ing it in conformity with some geometricaj
and brancbes. It was exteoded by tbe Caliph
'
rule. Cases also may be found, though rarely,
Othman, tbe successor of Omar ; he usiog hewn stooe
in which the geometrical method leads to a
and Indian teak wood on the work. At the begi0nin'
clumsy result, that can be remedied by freeof tbe eightb centt!ij' it was entirely reconstructed by
. hand drawing.
the Caliph Walid .I., the 7tb Ommiad Calipb, and 12th

DISCUSSION.
.
The Chairman, in opeaing the discussion,said,
3[r. Hankin's claim on their attention was that be
had discovered at Futehpur Sikxi, the Windsor of
Akbar and his son Jchangbier, in the bath-room
attached to the " Jodbbai Palace, the actual
<!raughts of some of the geometrical projectioos
OIl which " the Saracens " are believed by many,
and Mr. Hankin shares tbeir belief, to bare planned
the characteristic pattcms, or at least the more
elaborated of the patterns, known as " arabesques,"
nsed by them in the decoration of plane sur" Jodhbdi, was the Rajput PrineeCS oF the House"of Joclb:
yur, who married the Emperor Akbar, and became the
mother of bis son Prince Selim, better koown by his imperial
Mme of Jchanghkr. The Jodhbai Palace was sacri)cgiously
aad outrageously blown up by us as an experiment Id milk
Bary miniog in mo.

from Mahomet, who, ambitious to make it the most

splendid public building in the world, asked tbe com
temporary Roman Emperor at Constantinople to
supply him with Greek workmen for the purpose;
and the Emperor sent him 40 Greek aod 40 Coptic
workmen, with 40 loads of " Greek glass," beside
sih'er lamp chains, and other finely " arted " appliances, for the more decorous furnishing of the
Mosque. This same Walid had pmiously, when
about to build tbe famous j;lnll"-ma$jid at Damascus,
and the so-called " Mosque of Omar " at Jerusalem
[the kul0atu's-3akhra/l, or " Dome of tbe Rock "
of the Muslims, because occupying the site of
King Solomon's Temple], entered into a regular
contract with the Roman Emperor at Constantinople to supply biro; according to his need,
with /'Jy'a [also written ftes'jfisa, and /es/esa],
this Arabic word for " Greek glass," being none
other than the Greek word PsePhos "a pebble."
'The " Mosque of the Prophet" at 3fedina was

'

370

THE MATHEMATICAL GAZETTE.

EXAMPLES OF METIIODS OF DRAWING GEOMETRICAL

ARABESQUE PATTERNS.
By E. Hanbury Hakkr4 M.A., SC.D.
tbioxg the complicated rectilinear pattcim emplo 'ed in Samcenic nit thrrr
is a Qup which may be described as " gcometricaj arahcsqum," drawn by a
curious method which may be of interest to mathematicians.*
A " repeal " of a complicated pattcm of this class is shown in Fig. I.
ccnt.rc of t-he reyat is occupied by & siAtcen-rilycd 6tar.

The

A quarter of Bucll a

[<~aj ,_^K^Yau> "

)))"14'))7)))")')"/)7
!)'J-:?Yi)'72)j')'"{)),'")j)'")!'i)
=: (}t.,z!:.".!!!!}!"S::
a

'

Fla. I.

Blar Occypics c:tch of the four corners of the repeat. In the lower ]cfthand
quarter of the clrmring the construction lines are 8}1own, by which this pattern
may be easily c] mum. These constmction ]inc, Bhown as qinglc jiijcs, conuid
of certaiyi pulyyms. The came of each Ode of each piolygon is cFossccl
.

' An 3cc.unt rf nil' discovery M this mcth.jd, In; nndinS mo? of the crigiml construcfl m
j::m, was jmbjkhcci in :j parer "Cjn mite ])jx"cA;cr;c$ of tu3 3[ei')c4j$ of Dcsjpn cnqdrycci I::
,Yi;uh(pin('r!An Art " in the Jovntnt of Ge Q'oidV c', Ark, Ines. rol.'!!lL J'. all. A jnorc dc1?,iX j
Accctgnl i:. c-n ::le poInt of publimuon in the :jmoi:s og the ArMg.oqicai Dgmrtmem el lia
G0l'nltjicRt qf hidiai

GEOMETRICAL ARABESQUE PATTERNS.

371
.
by two putcm line3. The motb of drawing these pattern linm may bo
c.wilv seen by impcction. Those that enter tho construction octagom
mu drawn pnraljd to their radii. Certain of the pattern lines entering other
p(Argon3 are drawn in lino with each Other, and bo on.
'i'hc only diMculty 13 in drawing the o]ygom. One has to commenco with
tho octagona. To dO bo, proceed a3 fo]]ow8. The point B is the ccntrc of tho
rrpr.it. Divide the angle ABC into four equal parts, which are indicated by
dntlCil jjnc8 Do tho Mime with the angle ACB. Tivo of the dotted ]inw
intersect tit E. Then E 18 the ccntre of tho octagon required. Another
inhrstion occur3 fit O: Then the line ED 19 the radiu8 of the octagon.
Oihcr pdygons arc fonned by prolongim; the radii of the octagom, as shown
in tho Jcfthand part of the drawing. Tho exact size of the 16gon, which 19
the lmis of tho ]6-rayed stars of the tqattern, doc3 not matter. If GH 13
in.irkcd O equal to AP, then BH, or
ereaboijt% will be the radius of the
h3gon.
.\ curioiw modification of t]]i9 pattern 13 5]jown in Fig 2. The col'nci5 of
the rcpcu are each occupied by cjuartcr3 of sixtcen-r:ij'ecl stars, but the

I)))))
).))y/j:
;j"g

M.

J)

R
K

.\ N

A'
C

m'
FlO. 2.

';'.}i!,1r star has vani'9ied from the centrc of the pattern. A3 diou'n in the
l'nrtr !cft.h:;t;.t ."ut (pi the drawing, the construction lines consist d octazons,
"lp:'/l!f zncl iircguhr heptagons, bcsidcv the quarter cI the !6-gon. Conj-

372

the mathematical gazette.

mcnce by drawing tbe two construction octagom of the same size ag and
from the mme ccntrcs as in the previous pattem. But four Bides only of c.wh
octagon are dr:ium, as, for inmnee, CD, DE, EF and FG. Them draw 8
complete octagon of the smne size (as K), with lwo of its angles cc.incidin;
with the points G and II. Repcmt these complete ochgons at L and M. Jnia
opposite angles of these octw'ons, thereby forming oblongs N and P. Thcrc'bv

irrcEnlar hcpta"ons as A and B h.we been fonned as residual spacca Jlc.it

these irregujarj]cptagon$ at R and S. Having completed these construction
]ine3 the pattern may be drawn by the mme methoa a8 before, and nccch nq
further description.
A singular example of the ingenuity of the Arabian craftemen is shown in
Fig. 3. The repeat of this pattern is El sqmrc. Each of tbe four comm jq
g

)78
Flo. 3.

occupied by the qu,u1cr of a twenty-rayed star. The ccntre of the rcpc.it

is occupied by a twclvcmycd dar. Divide t.he nngle NBE into fire cqml
puts as indicated by dotted lines. Divide the nngjc NEB in(o three ccj:rj
parts. Two of (he dividing lines infcrscct at the point F. Then draw .7
qmrtcr of a 20gon ui;h radius BF and a dn'lrm;cm u it!) r:idins EF.
These largo polv:ons cjrcl'lal) by two Qjji.4 at G .ind //. AL JI they Iq ""'
a residual pjmcc. " In this 3fKICC draw ,1 hrj't 'Fnj) as rc',;nl.ir 7v3 pc'tiklc, I)r\ i:s;

GEOMETRICAL ARABESQUE PATTERNS.

373

tlircc of its angle touching centm of noijhbouring 8ide3 of polygom. The

hepl.igon thug drawn become3 part of tbe pattern. Contrary to tho usual
pr.icticc, no further we 13 made of the 8i(lc3 of the polygom in drawing the
rmt of the pattern. From centrc3 E and B, two circlm arc drawn which
touch each other at 8 point midway between G and H. ThCaO circles are
,hown ns continued ]ine3. Pair3 of pattern lines are drawn throtigh each
poi:it where theso circles intersect a radim. Inside these circleg slightly mailer
circlc8
ciro drawn,
whose 8ize
to be
essed,
whichside
are drawn
dotted
lincm poim
ar0 marked,
as ha3
indicate5"by
dots,and
od each
of theas
position
\\jlc7c these circles cut interradii. Putem firm meet in these dot3 a3 indicated
M P. Tho exact distance of these dots from the interradii doe3 not matter.
Nn:i1lcr circlw at R and S are ako required. Their exact size 13 unimportant,
.md 13 a matter of Euess-work. The point to be aiined at in n)akinf these
kuca$c3 13 that the pattern space diould be as regular as possible. I radial
hynljljctry 13 impossible, at least a bilateral symnietiy should be aimed at.
A feature of all these patterns 13 that two lines crow at a point, but that a
point 13 never a ineeting-place for several ]inw. Hence each line, after aiding
in limiting one pattern space, nin3 on to aid in limiting another pattern space.
Tim the didercnt parts of the pattem are interdependent. This appcaranc
d interdependence may be increased, with addition to the aesthetic e5ect, }
i:itcrhcing the ]ine3 as in the examples here draim.
Usually, ag in Fig8. 1 and 3, each pattern lino starting from any point o
tho margin of tho repeat may be followed in a zigzag course through t}jt
pi(tcrn till it rcachm some other point on the margin. But in some cases,
.1. in Fig. 2, mine of the pattern lines mark out closed areas. It is impossible
Int' mo to 6co from the construction line whether or not such cnclosurw will
kc: found in the completed pattern.
D! tho patterns hero described, Fig3. I and 2 nre from a dome in tho
.\!h:imbra (see Moorish Remains in Spuin ; the Alhambra, by A. E. Calvert),
micl Fig. 3 is copied from Bonrgoin, Les Elements dc l'Arl Arabe : le Trait des
/.'idrc!acs (Paris : Firmin-Oidot et Cie, ]879).
E. Hax"bury Hanktx.

SOME Di:.klCuLA DELi:A\LI.6\A.U UL.3lu.\j

AU

SOME DIFFICULT SAIZACENIC DESIGNS,

By E. Hanbury Hankin,
II,
A Pattern coNTAINwa Sevfn-i1ai'd Stars,
T}[e ideal aimed at in Sarnccnia dcsigm is that cnch " pUtern
Bpace " Miould havo either radial or bilnteral symmetry. It is Blll'-

prking that thi8 ideal 18 bo nearly achioved in a pattern in which tho

very awkward constituent of 8even-raycd stars ia conibincd with
octagojw and regular Bquaro-8h8ped oublines. This pattern 18 diown
in Fig. I. At fir.vt ghincc each of the pattern )1lcc.4 8cc]t)8 tc be
r_"

U4aAy-dkAty

,'A:rV^2t

RiTikSSEt
lV
Yiq. I.

perfectly 8ymn]ctricm. But on examining the trilobcd )(lcc8 of

which ono may bo 8CCll on each Ado of tho central point, it is CCl8Y
Li) sc-q thu the anglo between each two nd jncent lobess ix in two
jiMimccs it right angle, but t.hc third of Lheno angh:s is [l little htrger
f.hnn a right angle. -

166

'

THE MATHEMATICAL cAzE1"rF,

The illustratiou shows one " repeat " of this pattern. As the
illustration shows four square outlines each containing an octagon,
it might be thought that what is drawn is four repeats. This is not
the case for the following reason. Ab the ccntt'e of the illustration
is a c1iauloncl-s})al)c'd space. A quarter of a similar pattern space is
to be found at each of Ute four corners. The diamond occurs, in'each
case, with its long.axis horizontal. At the centre of each of the Bides
of the illwt.ration the diamond outline occurs again, bub with the
long axis vertical and therefore to that extent difkrenti from the
similar outlines occurring at the four corners.
A rule that is strictly followed in Saraccnic art 13 that every hand
to be decorated carries either one repeat or & whole number 'of
repeats of its pattern. Consequently, as may bo seen, tho pattern
we arc discussing is suitable for a s,quare panel.
.

-\"C\

PP
0

"'"--,. Rr

,y

C,

k , .. .H : :""".

: / '"""~~ ~

n'. :

,,.S""":,,. .-':"::":.".j{i'jj'::':
2

K O WP

. P

"" "'"".'\' A'

i O^"""""$ i

0
0

P
0

'j

7' '

',r'i '0
f'

0
0

'. ., Y\ T

0
0

1 :
t i '
W'

0
W'

'i

'

'D'

FlO. 2"
W

How is it to be drawn ? Suj'po'a we have a square panel, how do

we set to work to draw this pattern on it ? The method k 3howu in

SOME DIFFICULT SARACENIC DESIGNS

lG7

Fig. 2. One has to begin by dividing tho baso AD into three parts,
of which tho central part BC is equal to the length of the side of [l
heptagon whose diameter is equal to tho other two parts AB and CD.
Lacking any mathematical knowlcdg,c, it was necessary for mo to
find thi3 by trial and error. Having lound it, and transferred it OIl
to tracing paper, the next stage 13 to use it in covering the panel with
; lattice of ovcrlai)ping heptagoiw a8 diown in the figure. These
.orin tho chief construction lines for the pattern. Having drawn
these
be legin
tagons
and such
their diameters as may be necessary,
one
can
drawing
the of
pattern.
It 18 advisablo to commence with the square outline. Its position
in relatiou to the construction linea is shown in the figure md needs
no description. The length of the sick of the square 13 equal to the
distance from tho centi'e of a heptagon to tho ccntrc of the nearest
adjacent heptagon. Each side of tho squaro is interrupted, as at
E and F, where it reaches a radius of the heptagon. From the centrc
of thi3 heptagon at tho distance E draw a circle. All the inner
limits of the rays of the seven-rayed 8tar fall on t}li3 circle at points
whero it is intersected by radii.
The next pattern lino to draw 13 SR. This darts at S, a point on
the side of the heptagon which is obtained by drawing a small circicj
a8 shown and which needs no further description. The other encl
of the line SR is at R, where a radius cuts the circle already drawn
in the heptagon. This line SR cuts an " intcrradius " at II. From
the centre of the heptagon at the distance II draw a circle. All the
outer limita of the riij'8 of the stqr fdll on thk circle at points where
it is cut by interradii.
7he next lino to draw 18 NM. Ono encl N is at a point where li
rac,ius cuts the inner circle. At jr the line passes through ? point
where the outer circle is cut by all intcrradius. The line is continued
till it meets a lino now to be described.
'
Tho coiwtruction lines form a square space T. Round this space
draw an octagon. Each Bide of the octagon is continued in one
direction. It doc8 bo till it meet8 the line NM or other lines of the
8ame kind.
The pattern 13 coinplotcd by drawing, in appropriate ylaccs, lines
similar to thoso we have described.
The dmuing of thi3 pattern 18 obviously remarkitbly simple when
once the correct construction lines have been discovered. This cannot
be Bald of the "curvilinear arabesque "shown in Fig.3(p.lGS). Froni
it3 general structure wo must conclude that it, is, in sotno wav,
based on tho preceding pattern. But I havo failed 'to draw it with
8uch regularity as is presuniably attainable. My drawing, shown in
Fig. 3, 8how3 no very obviow faults. Each sevenpointed star
a'eam fairly sylnnletric31. The star is surrounded by seven shield8uaped pattern qmces. These 8hield8 ought to be of the saine size
and synuncti'icaL In thk I have not quite 8uccce(lc(|. The pattern
13 to be fcmncl on Plate III of Buui'goink Elcmt'nls dc l'url Arabtt :
Le Trail clcs Enlrtlucs (J"il'lljillj)itot el Cie, l'aris, 1879). !ktt. hk
drawing i's on Sc) small a scale that. it is ditlicult to know how Fir lie

lG8

thf m'nlE6[A'nc:Aj, GAZts'j:i'E

has ac'hierccl regularity. Neither docs it; u'rve to givo any clue to
his method of construction.

! 'I

-7/

))')r'Lir
N I

'\

'\ \

!=/i!i9

Ii

' \>
i

g'
FIQ.3.

A previous article by me on the same subject may be found Ir this

journal for 3.[ay, 1925.
E. H. H.

318

THE MATHEMATICAL GAZETTE

SOJIE DIFFICULT SAR.ACENIC DESIGNS. I],

A PAITERN CONTAINING FIFTEEN.RAYED STARS
By E. Hakbury Ha.17un

The dcsicm shown m Fig. I is unusual in the amouiUi of pattern tii,il

~

\)I"
Fig. I.

goes to 02c repeat, only one) coinpkto repeat being included in t],c
iilustratic'n. It is also unusual ir. that it includes fifteen-rayed stars

"""""""=":Est . "~-

(i:m"j:"""'1
"b .::<1'\ ;C-j>':,..
yj. , \

..r[

Fig. 2.

As indicated in Fig. 2, the repeat occupies [l rectaDEie whose

diagonal makc3 angles of 3Cl mrid GO dec:m with the 2djaccnl &idcs
One bcQus the censtruction ijne8 by drawing on such a rectangle '

SOME DIFFICULT 8ARACENXC DESIGNS

319

. of equilateral triangles. One of these triangles is indicated

,it.iice
loy the letters ABC. In each of these triangles draw & fifteen-sided
olygon. Dodccagons are then described whose ccntres correspond
0 the angles of each equilateral triangle. These dodecagons are of
.itch a size that (.he length of lbeir sides is equal to the length of the
.iile of a lS-gem. Adjacent corners 0[ t.he 15-gons and 12-gon8 are
joined by lines forming pairs of small triangles a.s shown at R. The
main construction lines are thereby completed. The pattern 18
,oiistructed by drawing two lines crossing each other through tho
ccntres of each side of each polygon and triangle.
It is advisable to begin with the lines DE and FG. These are in .
line with each other and each :jasses through the centre of one of the
sides d the 12gon, Inside the 12-gon these lines come to an end at
the points E and F, which are at a small distance, that must be
guessed, from the interradius SC. From the centre C at the distance
/{ or F describe a circle ; this is partly shown drawn as a dotted line.
All the other lines that enter the l2-gon from the outside terminate
on points on this circle at the same distances from interradii as the
t.wo lincG already drawn. From the point; E draw the line EK nearly
parallel with tbe radius CT. This line comes to an end whero it
meets the interraclius CV. 'From the cent.re C draw a circle intersecting the pointk. All the other lines that come towards the centre
;f the 12-gon end on this circle at points where it cuts interradii.
Jimi]ar]y the line FT is drawn nearly pare llel to the radius CV.
Other lines drawn in the same way complete the twelve-rayed star
of jhe pattern that occupies t.he 12-gon construction outline.
4
Ao1v let us go to the fifteen-rayed star that is to occupy"the 15-gon
cons(.ruct,ion outline. Begin with drawing the pattern line HJ. Thi8
13 not drawn in line with the line ML bub, passing through the centre
of tb side of the 15-gon, it is given a.n inclination that makes it as
much as possible a looking-glass reflection of the line FG. Tho lino
llj comes to an end inside the 15-gon at a, point near an interradius.
From t-he centre P describe a circle, which is shown partly drawn as
adotted line, passing through the point H. All other lines that enter
the lS-gon end on this circle ab points similarly near intcrradii.
From the point II a line is drawn nearly parallel to the ncighbouring
radiw and ending on the next interradius at N. From the centre P
at iiista';ce N draw a circle. All other lines that approach the centre
nf the l+gon end on this circle at points where it cuts interradii.
The remaining pattern lines are those that form small hexagons
that occupy each of the triangles of the construction lines. These
litic8 cross the centres of the sides of the triangles and are so dram:n
as to make the remilting hexagons ;s nearly symnet.rical as possible.
This pattern may be found in .j^$ ElCmcnts dc CAtt Arabe : Le
trait des Enlrdae, by J. Bourgoin (Paris, Librairie dc Firmin-Oidot
CL Cie, 1879), Plate 128. It is also briefly described in my paper,
" The Dm wing of Geometric Patterns in Saracenic Art ", in Memoirs
if the A7c/IcLcQlogical Sutuc?/ of India, No. IS of 1925, Fig. 39.
E. H, II.

,'/ !'""\J TI1

Fig. '

Oblong of 60 degrees.
k

Fig. 2
Panel witable for three repeats of 60
degree oblongsD

,)')-" )) )

Ii9'.
XXX'('K

Fig. 3

Pattern of hemgotw in contact.

a,y""
(-}-1.':)-'|'

'

:Cx

(-!:..j).i!Ajaxx
Fig 5
b

PMkm drawn in pempective.

"':1(,
'". ",(,!",1\'"!'""'
,:'1,()" /_/1
/

\\\r

(ii |.'",';J'":

"ig'.')'))i'i"Y'

'(

)Yl)z

d
m

Fg. 6

Pauern of hemgon* and arrjw.heads,

m

rig. 7

C
He&gonm uttern from FhpurF4pin.

H~ & L O. VA

-.

"i

""'=" """K!>1""""""A " "" """""

.,- ----"b,.,!!j,,!:S... ~
< "
? ""' )Y "
0

"I-

"

%%%

m "" " "r

I % I J
%

pP

%%

"

i '"^ I

.^

~">'e'i:*: i-<"." >

<

&,2?;*sg>" <"" "

.i"-..::j2:.!4a-":"'Kt[im ./ I _ f

><:::,-..1.,::>--.., ~ r "1 i

G
CHexagonal pattern from fl tomb near helhi.

Fig. 8

j:'Y::')|:"{!:'"%:!"\\,,i):"(:j'$:i,':'{)::)):'v::j 7' / \ '\

')4)\:l0i('!):0')\:)(:,)':j);\))))Ii'I))'('\'){'":":
,I

,;N'iy'i' ,t[,mi':,:"i(\ 1\,,,.'

'

-'",'i \i/'w \s\i' "\,i" g ",1""

Pattern of krge and 8ljl8jj hexagona

Fig. 9

I)

B
t0
0

, - /,{G
\.
I--

Ei

C jr i- ,

, ," '"

, ;J^'

jjj'y" "
' 'cl' '" 'i"+' s%!|iSi$
y'

"' '? Y "

\,

___Gl "A"

/ 91) <

Fig.1o^, d" O"j /"

Flg.11k

/\

^ fig.12S^Y^XA

Method of clrnwing octstgom PUtern fonned by ocuigons

in ccmtru't by Bides.
in contact by 8i(le8.

Methnd of drriwing OCtRROIW Pattern formed by octnt

in comm by their angles.
in contact by Angles

CTY17
t--:--t"

ZKC-!

.K L,!"

7,{j:))t>\ j)j,
,,..,,G%klZ'iD ,,.. ,:=ii
l',-'r\!

.--..-I"*":::::::

i) jj6'

PMtern of mperpoxed
oct!lgon8.

j- j"/t\"" m=7

i=." """, ,\1= )

,,,,(

,,,,A-"-

"'jYKy' """"-

lx

-L^-L-

,X
)
0
^

F,g.16V\/\/A

Method of drawing

1'Mtern of oculgonA

ovor1nW)inR octagonm

over] tipping by two 8ide8.

Method of drnwing pM.tern

of Fig. ]8.

Pattern derived from ochigom

in contact.

Hklio. S. I. O. Gmmn

Plate ITT.

))')))
~^^ ))J

C
Fig. 19

Putem of tre}li8 8enen in a window of the Dafter Khana, Fuhpur-Sikri.

/"\".""

7~,

bb~

Q
V

;1 "\

"K.

1""4

\I

L ^'

i"

r""i\

"-X-

K.j

a
0

"4

"-" \j-

" "" J) ';') ,>,,,. ,,

G

,7---1 7-----..j,./ LLLY

V
, r ""

"W

^."

\\ ,,,/

Fig. 21

I XI

Fo. 20 Conutniction of an octagonal star and

octagons of two sizes as frequently

%
4

XL I >/

V I kt

A
Octagonal pdttem u8ed in inlnid mArble
in the tomb of Itiimdu-d Daula, Agm

found together in octagonal patterns.

2'S&xi"

A-"------. I C ' !,
0

A-------K,

p
P

I 'P

-- T

ft- i - - 4 -

m-- t

I l"1"',j'!---+-

ix %

,. " i

Fig. 22

I
I

'

-------- :/-zzi-2': :, -! i '

m '

I
I

----Wu

jg--l j)"' I"

,,....CS.l%S.5
)XJ>

Pnttc'm of pavement in the central

chamber of the Taj Mahal.

Pattern from the Court of the Lions in the ;\Iharnbm.

Plate I \'

)"))))))))I),'jn
""\ji' '"i" ""\,.r"' '"P-a- -\'

,LjpL-m

'

7 "

__ ..')\, ,_> i:,

'

\""' '

~%

'

N!! "-

. "

"

/' "

I))
"" %L' j1

\
p

-- """ """{ " )""' ""(' L Kp'::" \ / ,

"

A/ G) \"^j

') J)"",')'"") ,) j))):,.',')j'x|,:)),)),),))'))}v l-i)))'-)

~

""'\
Fig. 24

/^"j

>F'F7~TF}'"'
I

""

Pattern derived from overlapping octn'()nR, from the screen in the toinb of Itin)n(hl-(l ]hiuk, Agra.

))))')"))('1

.,..))'
Pattern derived from octngom in contnct, from tho vedibulo of Akbarb tomb Ill Siknndruh.

PLATE V

r
t

)
i
,,

"x

r"

rp"
4

.,
-

"t .

.....

.,.
....

"

,"

i
\

'
t

"\

/\ "

""'\

t ."

">

,t
rf

a"

.-.

~ /\ s

""-> <-\jt

, ,' ,,,

t
,
%

,
g

)'
)J) ))j)'
\

j t~

'

C-\ v>'k"C T-l

't'S,?YL,<\/

---

r1GjYTAr\)

---

Fig 27

"

PUtem in mom ander the Khw&6uh m

the mpujue M Fathpur-Sikri.

Fathpur. Sikh.

Fig. 29a

,m,

/'\>

sgmg)

PMtem of Gb screen on the outside of

/ '"y'" '~J

....

RMiE

Ng. 26

">

AJ" \J9\
^ ~/ Q
., -L

'\."

i ?x?i S,"\/"')---. .- --"'

<X><Y">d '
/ '7 \

\, "

"y"

De8ign from the Taj Mahal Hotel,

Bombay.

~t^
Aralmquo pattem d on QcWons in comm.
8

77-:)LA!a
!"' ""-:\, ,/

G'

ir

".

:
n

""""""'";"::},/ "

"-. ,

U!j":""'"""""'7

,~"'@i:\:""-i""i;:::"< ^

6,---('""")":/":"" -- ):S,/n, ,'dP

c

Fig. 29

":<S"'j "'

Q nq
ArUmquo prtttern bivucd on octngons and
irregular cnhl9Un&

'

j'j 4TF' VI

"

---" Tt

"'

-"-

--

-s:

/' '\/ \

,,,/ ". .'.""

"_""

'/. !'"E"'. '

h.

\,
,\

.'

\
.

', I

"'"" "" '"

'

.
L

'

-U\"""j.-""" " " ,'\

L \

.t-" t""" - /\ -a

.,)))'"")),)""),-.!))))|.))'"1))"_"
-I
jl
/

-.
-.

'^:,1

.( ,

'j,

" ---"'M/u""'

! ^:,jj" }C;

""

/)1:"),")Ag!
b

a
Fg. 30

Aralmqae pattern briacd on dodc~n8. squnres And equil ateml triFLnkk8

\ 'I /

,.
1'\

-JL-

|Ay>1?k',l^u

i 'y. """ """"" ""'.

_ -7 K

Al

i"

~'~

'"]"^'"1':
\Zic's""' K
I

"r

JN

\'

It

' --)\) ""jjj

Fg 3

Pattern bmed on c|cMiec&Xnnq Kelh4r8ted by eqni lnteml trinngle8.

'""I))'))))))")
-, L ~ }'P"

/ )Y " " "' "V

Y\.

\1

! -J

m"
)i

>"

,,/jG

p -

a
Fig 32

A mhmque putem lmed on deeaon8 &nd pcimgonz

g}:ij
'^'""'"
""'
'

- ~- X"

"R\-AJ

g-i- -i-- y-;>

^,.,r

:M 2><

9!1./'tj"~jQx
"" f "" "'--- - ' - ' :"K"QX"'-7
,L:;:,
// i .i i

g. 34

He" 0/

AmbeMlue putern based on 8 14-gnn, (rnm

tho enlmnce to the luonue io

Fg jj

Border of hrkc jmnel in Birlmlb

hcmw At Fnthmw-Sikri.

i'mhpur-Sikri.

'i""\ : K- ' T" "" <Z " ~,^,4\,

/jv/'

'L

: \~L
.' -~=-J ,

/ Is

D ( ;) , --

(" \ /,

^Y

"'

,<" ' :. ' . jk

..

-. 7_ _

T/'l

i-" / "' ' A, \ '

%
..'

>1

\
L.b
r

'

'/

L'

"' . ."

/'

. ... , ,...':..>
'

'5

-4'"y

L-dr"

. .. ..
.f
'

tzs:
( <7: "LA

^-"" \'

'

. 1/~

L '

4 . - '"
..-' '5
.. j,.-"'"
.

' '6

m. .~1 -

a
Fig 35

a
r
['8Ltcrn lutNed nn 14-gcpn8 md twinned

'

h('ptngnnk.

"

""" :

"\'j 7!>, /'\,Ar"'

,, ,' ' ,1) j;t "j-f ,

7A<" m ,/>

/'"' "i,ys
/

i/'j

: """"": J j,;/ " .'"".

-:,:1)")), 'i 12 C
"(..-...:y\?.-. j)kjgq'*
'""" /'

/_4Y"LA< \\/

:,:,,"" -" ( --- ' - )- - -rtzjj\/4/

E

"

/ Amtmquo pMtorn l~d on 16-gnns, 12-gom

and hepctignnm

,a

7jE"na-i'i?"" '))'"'",t'j
(':1"|'>
-C)
r

JEiL:za R=yY"t'yzma- Asixq:3\

d-t"Anu_m w'

I-l'

""i?)ij!jjjl)s7tt. _.

~^

Fg

36a

L-

. "~=im

'

\ _

v &

)j g(plb8, 12 gorm

wncl tit'pLng 'tbh.

,7 ,I), _') {i?Q?j2)ijj ,T,,

-. .

i"""' / "A \>,

'

ri

i'

A"k x b

A \AJ<N
\A '"' """>cur
7'j> :m: O """'"'

"E')'"^"
'

C > m""""|")|(i" \ / ">

>'t4,} Y

',L!:c."
.

"r 4$

\ ^c7"

^,:Znl3 \/\ ^/
"

^ J

>.

< , F!\g7?<'
" Y y "
"~1'\
Fig. 3t

' '

'i

AmhcNque pnttcrn brlRcd on dccA{y)n& nnd

8pin(llcshwpd tigurcs.

Jj

'"-"

Pattern from tho mflit nf an mb in the

Hakimk bath BE Fathpur.&kri.

.,.

~"""'""'W~j

r,

, )J'I /,, """

t

)b-:.::!:j),' )c-:(

J)
Fiq. 39

Ar&be8que pattern containing fifteenpointcd Atnr8.

.-

4/ A"? ^ C,U9<Z: ,

i ) )gAxjg|
y"y'".,z:,~ _

Fig. 40

Arabesque pattern conWining 18-pointed 8tflN.

>"\

-< " " "T"" ""7~

\Ar
7\ '1""t"""/ :L<zs:kr
'r-L>m )
n t l^

"7--/
'i

^. "jm',

:"
0/\

14 ',','wm m

:t>3xx

ifT" m+1"+iNX%FFK|b'tm -
g 1Llj'r 4/&AK:
tt"cj w\^V_\i,~r ^
i

LN
"m=rr '-i \m^trjL

----....,, )ji::='":E,~=-

l.

Fig 4

Pdtem containing Reven-pointcd dptra.

< ~6

'

":', ."

)))
\

3
r G, 4 2

~'>>

\C'/

)6-.'-)9)-|\
Decagon and a demgon with two

triiingkw cat out'

\'\/ ,/'

"

Fig. 40j '\ )//

' E'":'"'""Oecor8tion of a half-dome in the

\1 /

S'jnom M&b&], Fathpur-Sikri.

Ikwign from li domo vin l'hLhpurSiki'i.

v

,jii))))).,j|'-:'j.,,,
L \ f'iagmni of the CC .,,.

\.

,,

"7
stnicturc
of a
A~ \\ dome
of
inereas-

//*6k4tF?gilu'cuL'vjlturc"/

' %
/

,(NKMguYs>c~

'"CJ:"
"
,"xgj%>" '\/j"'
"""-L~ ^~,

' /
CV
' " Fig. 48

,k-

k/~k_/ \,

"

Dccomtion of lb hulf-clome in the

Agm Gate of the outer wall of
Fttthpur-Sikri.

pr

-76"'1
} jjy]),,,,,,.,,,,,,,,,,,,,,,,,,
EJ F

""

'\ /

Himn JJinar. F:lthl)ur-Sikri.

xlll'llb()llnting the clll)()hl OIl the