Islamic and Indian Magic Squares.

Part II Author(s): Schuyler Cammann Reviewed work(s): Source: History of Religions, Vol. 8, No. 4 (May, 1969), pp. 271-299 Published by: The University of Chicago Press Stable URL: . Accessed: 27/10/2012 00:55
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact


The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to History of Religions.

Schuyler Cammann




We have seen that in the Islamic world a magic square was far more than just a balanced array of numbers so arranged that it could produce an identical sum from each row, every column, and the two main diagonals. This was only a primary condition. The ways in which the Islamic magic squares were made, and the subtle interrelationships of the numbers within them, conveyed complex religious meanings. The same was true of the magic squares in medieval India.1 However, as the symbolism expressed
1 Very little has been written about magic squares in India. The only early source is Narayana's Ganita-kaumudZ, A.D. 1356 (hereafter known as G-K). This was republished in The Princess of Wales Sarasvati Bhavana Sanskrit Series, No. 57, edited by Padmekara Dvivedi. Part II of this, containing the chapter on magic squares, was published in Allahabad in 1942. I also have a microfilm of an incomplete MS containing this section: Indian Office Library, London, No. 596B. (A similar MS is in the Indian Office Library, Cambridge, England, No. 77.) I am grateful to Mr. D. Joshi for helping me by translating salient portions of the microfilmed MS in my possession. (In the numerous cases where the written explanations were inadequate, incomprehensible, or simply not given, I had to discover the construction methods for myself, by working through the squares to find out how they must have been made. As usual, there were many garbles; the method for resolving these is described in the introduction to Part I.) Otherwise, there are only a number of brief descriptions in the short-lived periodical Punjab Notes and Queries, I and II (Allahabad, 1884-85). The Indian Muslim squares, which belong rather in the Islamic tradition, are briefly discussed in Jaffur Shureef's Qanoon-e. Islam, trans. G. A. Herklots (London, 1832). 271

Islamic and Indian Magic Squares
in the latter was dictated by ideas and beliefs from Indian thought and the Hindu religion, the squares themselves were inevitably different in form, and often in their basic construction, from those that were made and used by the ancient Chinese or the people of the Islamic nations.2 And yet, some Chinese magic squares, and perhaps some Islamic ones also, seem to have provided the stimulus for later developments in India. In spite of the absence of concrete information from India-or Western writers have asserted perhaps because of that-many that magic squares were invented in India. When other authors pointed out that there were no factual grounds for such assertions,3 the pendulum swung to the other extreme, and some people claimed that India had nothing to do with the early development of magic squares.4 Actually, it now appears that the first definite magic squares come from China, from about the fourth century B.C., and many of the significant later developments were made by Arabic and Persian scholars, working from the tenth to the fourteenth centuries A.D. However, there is no doubt that some medieval Indians were deeply interested in magic squares, and they as we shall appear to have made significant innovations-though, the full extent of their contributions is still unclear. see, Probably the Indians' interest in magic squares began when they learned about the old Chinese Lo-Shu square of three. (See Fig. 3a.) They could have obtained this directly from China, either through Hindu merchants or Indian Buddhist missionaries returning from visits to China, or else through Chinese Buddhist monks who came to India as pilgrims. Alternatively, they might have acquired it by way of the Tibetans, who had been under strong influence from T'ang Dynasty China and had taken over the Lo-Shu for divination.5 Since the Indians had no feeling for dates in general, and never bothered to record borrowings from other nations, it is now impossible to trace how or when this happened. As we have previously noted in discussing the earliest Islamic set of squares, it is remotely possible that Indians might
2 The Islamic magic squares were discussed at length in Part I of this article (History of Religions, VIII, No, 3 [February, 1969], 181-209). The Chinese magic squares were briefly described there also. For fuller information on the Chinese squares, see my previous articles: "The Magic Square of Three in Old Chinese Philosophy and Religion," History of Religions, I (1961), 37-80, and "Old Chinese Magic Squares," Sinologica, VII (1962), 14-53. 3 See William Ahrens, "Studien iiber der magischen Quadrate der Araber," Der Islam, VII (1917), 217-20. 4 See Florian Cajori, A History of Mathematics (New York and London, 1926), p. 92. 5 See S. Cammann, "The Magic Square of Three," p. 75 and n. 131. 272

) The Dudhai square was found carved on the underside of a lintel from the collapsed doorway of a shrine known as the Chota Surang.Vereinigung. for the Year Ending 31st March..6 However. (If the doorway had collapsed long before. XIII (Leipzig. which is supposed to have been built in the first half of the eleventh century A.-The earliest Indian magic square compared with an Islamic one: (a) the magic square of four from Khajuraho. Examples of it were found by archeologists in two places: the first at Dudhai. was found in a Jain inscription that is assumed to date from the twelfth or thirteenth century.) The other example. it is possible that someone might have inscribed the square on the fallen lintel as it lay on the ground. but we lack any direct evidence for the use of magic squares in India itself before the eleventh or twelfth century A. Ic) with its sequences written in a variant order (3. 7 See the Jahresbericht der Deutschen Mathematiker. 1. seems dubious. 1. Hindu and Buddhist Monuments. 1. and even this earliest evidence suggests a counterinfluence from the Islamic world.D. Northern Circle. 383. 273 . the Islamic scholars were fond of experimenting 6 See The Archaeological Survey of India: Annual Progress Report of the Superintendent. and the dating 7 12 1 14 2 13 8 11 16 3 10 5 .History of Religions have had some part in the first transmission of magic squares from China to the Islamic world. (See Fig. 4. 3.7 and in this case the dating is probably more reliable. and the other at Khajuraho. but he stated that the square must have been carved before the temple was erected. in the Jhansi District. from Khajuraho. 1916).D. (c) the basic Islamic magic square of four. p. As we have already noted. commencing at the upper right corner in the usual Islamic fashion. 2). 1916 (Lahore. The archeologist who found it described the lintel as fallen. 1904). But is this square actually Indian? If we rotate it 180? we can see that it is the familiar Islamic basic magic square of four (shown in Fig. The earliest known magic square in India is a square of four. because when the lintel was in position the square would have been concealed by a supporting column.9 6 15 4 4 15 6 9 8 11 14 1 13 2 10 5 7 12 6 4 15 3 16 9 5 10 3 16 11 8 13 2 b 14 1 12 7 FIa. (b) the same rotated ]80? to show its relationship to the next. this argument is questionable.

I (1884). above.D. 2. this. and G. "An American Puzzle. 2-9. was apparently merely another Indian version of the latter. as it may appear tilted. Narayana's favorite magic square of four might seem to have nothing to do with the Islamic basic square. However. 6-13. and it remained in use for many varieties of magic charms down to modern times. altering the sequences to get new variations. too. However. 4 5 8 14 11 2 15 10 3 6 15 10 7 14 11 13 12 1 FIG. even though the Khajuraho square may have been a borrowing. yet actually. Grierson. (b) the square from Gwalior Fort. 16. 1. 2a). 9 See n. Its numbering begins at the upper left in Indian style. and he later repeated this in other contexts. to the extent that later examples may not be easy to recognize. although sometimes its short sequences were augmented in order to produce other sums. as well as having augmented sequences. 1356. its general form apparently appealed to the Hindus. Narayana seems to have prized one particular 1 4 a. 4-11. illustrated by a number of variations on the Islamic basic magic square of four. in the usual Hindu fashion. 14. because of beginning at the opposite side). 5. for full reference. 8 13 12 7 6 5 16 9 16 9 3 2 b. 10. An entire chapter of this book on Indian mathematics. Another form of Narayana's favorite square was discovered in the last century in a ruined temple in the Gwalior Fort.-Indian magic squares of four: (a) Narayana's favorite square. A. At first glance. written in A. is devoted to the construction of the squares. X (1881). we find double half-sequences: 1. This is a variation that the Muslims themselves sometimes used. 1017. 7. 274 . 90. 8-15. entitled Bhadra ganita.. The base square is not always immediately recognizable. 12.Islamic and Indian Magic Squares with their favorite magic square of four. all commencing in the upper left corner. but if we trace out on it the pattern of the Islamic square (in reverse. etc. reversed. magic square of four above all others (see Fig. 537. items 462. and this seems to be a good example of that. or rotated.8 The single authoritative source for early Indian magic squares is Narayana's Ganita-kaumudZ. 3. bearing a 8 See Punjab Notes and Queries." The Indian Antiquary.9 This begins with a long discussion of magic squares of four.

That was the reason for the popularity of this type of square among the Hindus. Part I (Calcutta. 3. the Indians made up for this with other kinds of squares. Though this Gwalior example may look quite different. In Part I. "On an Ancient Indian Magic Square. 3a) since this square was much 47 58 69 80 1 12 23 34 45 57 68 79 9 11 22 33 44 46 8 3 4 1 5 9 6 7 2 67 78 8 10 21 32 43 54 56 77 7 18 20 31 42 53 55 66 6 1719 30 41 52 63 65 76 16 27 29 40 51 62 64 75 5 26 28 39 50 61 72 74 4 15 36 38 49 60 71 73 3 14 25 b 37 48 59 70 81 2 13 24 35 FIG. considered the magic square as a universe in microcosm corresponding to the greater Universe. but found no magic squares on them. in a manner that resembles the course of the threads in weaving.. p. Probably this ingenious method of construction was discovered through a careful analysis of the Lo-Shu magic square of three in its inverted form (Fig. every pair is seen to alternate direction. ). The more recent literature contains many references to "the magic square on the gate of Gwalior Fort. See also the Brihadaranyaka Upanishad. 66. then. 8. 4. these small symbolic universes resembled the greater one in their tightly interwoven construction.D. 275 . Cut in a Temple at Gwalior.l1 Thus. 1842). 2b). XI." Journal of the Asiatic Society of Bengal (N. 10 Captain Shortreede.5.History of Religions date equivalent to A. and the Mundaka Upanishad. I have personally twice climbed the citadel at Gwalior to examine the gates of the Old Fort. 292. it is the same square with its corner blocks interchanged diagonally.S.-The Hindu continuous method: a continuous magic square of nine contrasted with the Lo-Shu square of three (inverted). III. 1968). When one counts out the numbers in each in regular order (1.. 2.). 148310 (see Fig. even if their construction pattern may have been borrowed from another culture. we already referred to "the Hindu continuous method. In this respect. with somewhat less versatility." If the Hindu magic squares of four seem to echo earlier Islamic ones." apparently referring incorrectly to this one. too." an example of which is shown here in Figure 3b. Both of the squares in Figure 2 also embody another pattern which caused them to have a profound significance for the Hindu people. II. For they. 2. Unknown India: Ritual Art in Tribe and Village (Philadelphia.3-8. 3. these two squares were "typically Hindu. and they conceived of the latter as a vast fabric woven by the gods. 11 See Stella Kramrisch.

his squares look highly sophisticated. dropping one cell for the break-move after inscribing the base number (n) or any of its multiples. although we 12 In the 0-K (1942 ed. to form a horizontal cylinder around which the numbers can wind. There is only a slight deflection as the numbering drops one cell for the break-move. at least. 393. center) they go over the top to reappear at the bottom. and thus learned of the concept of "perpetual motion" within a square. 3a). skipping a number at each breakmove. both made with augmented short sequences. we shall see that there is reason to believe that people in India must have known this method since the early thirteenth century. and they cross over the right edge to reappear at the left. It seems quite evident that Narayana tried to render them especially complex to conceal the relatively easy mode of construction. who might then have been inspired to see if the same process could be extended to make magic squares of larger numbers. As with the Persian continuous method. to resume again. We shall also see that. One must imagine that the top and bottom of the square can meet. because the first examples of this Hindu method do not appear until 1356. (These two formations could be combined in an "anchor ring" or torus. and from the final break-move-after the highest multiple of the base number (n2)it returns to the 1. there is a magic square of five and one of seven.). even there the effect is still sufficiently striking to have caught the attention of a keenminded investigator. and ends in the center cell of the left edge. In this method. even though they are more than a century later than the earliest known Islamic continuous squares. and proceeds upward to the right. the numbering commences in the center cell of the top row.12 and. Although the same thing happens on a very small scale in the inverted magic square of three (Fig. to make a vertical cylinder around which the numbers can spiral upward. before commencing the next short sequences. each of these squares begins in the center cell of the right edge. for a potentially endless round. the progression of numbers seems to be continuously flowing. p. It would be tempting to assume that this might have occurred after some Indian had seen a Persian continuous magic square. As a further peculiarity. the numbers may be seen as marching upward to the right: from the 1 (at top. suggesting a considerable familiarity with this method. However. while the left and right sides can also join.Islamic and Indian Magic Squares favored by the Hindus and they doubtless subjected it to a thorough investigation.) When one can visualize this concept. when they were presented in Narayana's book. 276 .

which involves constructing two preliminary squares. passing through the middle number." during which the individual dies. or-in another view. nor do they appear to have been aware of the special properties that caused other peoples to regard this number as a highly significant symbol. to be reborn back into it. In the Islamic squares made by the Persian continuous method. This. for another succession in a potentially endless round. 277 . in both of which the method of procedure was somewhat obscured.History of Religions lack any traces of the initial stages of magic squares in India itself. This time. He was far more interested in a still more characteristically Indian technique. to be recreated. this phenomenon is more obvious. the middle number is not even remotely involved with the renewal. However. clung to the Persian continuous method and disregarded the Indian one entirely. or boundary of the square. the actual magic square. we have noted that in these the final break-move leads back to the 1 to resume the cycle. the last number falls at an outer edge of the square. That is perhaps why the people of Islam took so little interest in the Hindu method. however. The Muslims of Northern India (and modern Pakistan). and we cannot help wondering why it seemed to have comparatively little appeal for Narayana. after they had learned how to do it. Since the numbering begins at the edge. we do have some reasonable evidence from other countries which seems to indicate a fairly early development there. leaving the world for a brief time. 13 See previous note. was also a typically Indian method. To return to the Hindu continuous magic squares. They valued the squares made by this technique for quite another reason. the Hindus do not seem to have been at all interested in the middle number of magic squares. and from there the final break-move leads back to re-enter the square at the opposite side. involving a much longer time span-the world itself is destroyed. as it takes place at the center of the square. For he only showed two examples of it. and only resumes again by re-entering at the opposite side. In the Hindu continuous magic squares. then using these to form a third. this figuratively unending-though recurrently broken-procession of numbers enabled the Indians to see each square as a perfect representation of the Hindu (or Buddhist) conception of Life as a process of "Endless Becoming.13 then hurried on to another topic. as we have seen. which thereby acquired a special significance. then. for example.

then complete each column by finishing the other sequences. 10. to place the sums in the equivalent cells of the third square. but instead of using the zero he took the 25 at the opposite end of the scale. he would set down in its center column the first five primary numbers. after which the second. Finally.) The Hindus used this technique of combining two squares to form a third for making magic squares of other. thus the root square was composed of the numbers 5. and write the same five digits across the middle row (from left to right). 2 . 1. set down in the same order. and the result was a magic square augmented by 5. which became the magic square proper. odd numbers... 4 5 5 1 1 2 1 2 3 2 3 3 4 4 5 15 20 0 5 10 20 0 0 5 5 10 15 10 15 20 14 10 1 22 18 20 11 7 3 9 24 5 21 17 13 2 3 3 4 4 5 51 1 5 10 15 200 2 10 15 20 0 5 2 23 8 19 15 6 4 125 16 12 FIG. 4. 5.1). moving from top to bottom. made by flipping b upon a and adding the numbers in the corresponding cells. a man would begin by making three 5 x 5 square patterns. (b) the root square. he used the same five primary numbers in the same order." 0. However. 20. or "root square. except that this time he used another set of numbers: the "root numbers. p. 389).) Then. (The general effect is as though. the construction and final form are identical with those shown in Figure 4c. he flipped the second square over the first-left column to the right as if closing a book-and added the numbers in the corresponding cells of the first and second squares. Taking the first of these. higher. advancing from the right edge to the left as if passing around a vertical cylinder. after which he completed the other four columns. To complete the columns. 15. 15. 10. he took the second empty square and filled it in the same manner. 278 . 25. 20. 4.].-The Hindu method of combining preliminary squares: (a) the primary square. Then the second 14 Narayana gives essentially this square (see G-K [1942 ed. one started the sequence one cell higher each time.Islamic and Indian Magic Squares This method is probably most easily illustrated by describing how the Hindus constructed their magic square of five. as if he were proceeding around a horizontal cylinder. then back to the top." could be filled with the set n(0. in moving to the right after leaving the center column.14 (See Fig. It was only necessary to take a primary square and write the numbers 1 to n down the center column and across the middle row. To do this. n . using the numbers in the middle row as guides. (c) a lozenge-type magic square of five.

). but I have never seen this used elsewhere.. Add. 279 . shown in Figure 4c.15 The common Indian magic square of five. and the even numbers separately clustered in the corners. and it resulted in very 15 Of course this can also be accomplished by initially writing the numbers in the second square in the opposite direction: working toward the right in the first square. But. with the odd numbers all gathered in a lozenge. one would expect it to show some indication of the principle of continuity. in general.) Another. and that the other numbers follow each other by moving two cells up and two cells to the left. in a kind of "extended knight's move. so it seems safe to say that. since the preliminary squares of which it is made were composed with reference to the idea of the double-cylinder or torus. Magic squares of this type may therefore be described as "lozenge squares. we observe that there actually is a break-move after each base number. p. 375-78. and when we study the relationship of the numbers. down through the centuries. this mode of progressing is not a general one but is confined to this square of five. pp.History of Religions could be faced down over the first. shows an ingenious way of making lozenge squares by another method." However. as each successive odd-number square made by this particular method has still more extended moves. 48b. or diamond. as Narayana demonstrated. where we find "lozenge squares" of five and seven. more striking effect of this type of construction-using preliminary squares of this specific pattern-is that the numbers in the final magic square assume a particular formation. continuous magic squares for odd numbers.16 (The formula for the length of the steps. all lozenge squares are products of the method of combining two squares to make a third. as we shall soon see.1)/2. the Hindus had a specific symbolic reason for insisting that the second (root) square should be flipped over to face down upon the first. 16 This contrast can be seen in the British Museum's Persian book. to place the sums in the equivalent cells of the third. 7713. 50-51. it was also extended to make squares of "doubly-even" numbers. 18 See G-K (1942 ed. 17 Ibid.18 The latter required a somewhat different arrangement of the numbers in the two initial squares.17 The method of combining preliminary squares remained the favorite Hindu technique for constructing odd-number magic squares. pp. at the center." and wherever they appear we can be reasonably sure that they must have been made by this method of combining two preliminary squares. is (n. Yet. and. might on first thought be described as "static" in contrast to the obviously active. and the numbers in the corresponding cells of these two added. and toward the left in the second. both horizontally and vertically.

p. there is a second type in which the upper right and lower left quadrants consist entirely of even numbers." I failed to realize that this magic square was an Indian one. mentioned in n. These patterns which represent the contrast. This latter method must definitely have been used to make some other magic squares in this MS (see n. We have already seen that the earliest complete set of magic squares in China. The actual construction method turns out to be far more simple. although the method of making a magic square by combining two auxiliary squares was not made public in India until 1359.19 These characteristic patterns make the Hindu doubly-even magic squares just as distinctive as their "lozenge squares. Instead of lozenge effects. This is quoted by Li Yen. it is still probable that the method originated in India. I naively presented an analysis of the deployment of its numbers as an explanation for its probable construction. or balance of the odd and even numbers serve to emphasize the role of the magic square as a symbol of the coincidentia oppositorum. a magic square of eight. p. it seems that he must have been alluding to this particular form. 22 When I was writing "Old Chinese Magic Squares. because al-Bfini refers in one of his books to a "method of the Indians. 23 Yang Hui frankly admitted in his introduction that he was merely passing on some mathematical curiosities which he and some of his friends had gotten from older books in their private libraries.. too. 61. then. and yet it was admittedly composed from earlier books. 21 It is not always clear in the British Museum's MS (Add. though the absence of easier. while the other two quarters contain the odd ones exclusively. Then. some may have been made by the method of combining squares. 20 A study of basic patterns can be most useful for classifying or identifying unfamiliar magic squares. the topic of magic squares formed part of an esoteric 19 Examples of both these types are shown in ibid. we find squares of eight in which each row has an exact alternation of odd and even numbers-or even and odd-while in the columns the same kinds of numbers meet only at the center. 25.22 This particular situation provides a good illustration of the fact that priority of publication does not necessarily establish priority of invention." and can serve equally well as a clue to the construction method. 17. 7713) whether all the "lozenge squares" were made by the second method. 378. of the first of the two types just mentioned.21 However. of 1934).Islamic and Indian Magic Squares different patterns of distribution. partly because it had been tilted 90? to rest on its side. was not published until 1275." and although he was not very specific. now known. below). above.23 Furthermore. appears in Yang Hui's "Chinese" set published in 1275. odd-number squares made by this technique shows that it must have been foreign to China. 280 . Chung-auan-shih lun-ts'ung (pai-hua ed.20 The earliest known examples of the lozenge squares appear in Islamic books of the thirteenth century. Therefore.

though it must have been known before 1200.). which must have taken a long time to work out. 900. 7713. Narayana himself lists a number of sages or gurus. we see that this technique must have been primarily an Indian development. 281 . Science and Civilization in China. 25 The unit-centered magic square of nine in the British Museum's MS Add. The zero first appeared in Southeast Asia early in the seventh century. and possibly the unitcentered squares of five on the same page were also. who had not either happened upon it or been shown how to do it. see Joseph Needham. And actually there was a way in which someone could have quite naturally happened upon it. while the thirteenth century Persian manuscript in the British Museum contains quite a few odd-number magic squares made by this process. nor by means of exceptionally brilliant mathematical analysis. 24 This is mentioned by Padmekara Dvivedi in his introduction to Narayana's G-K (1942 ed. 1959). In contrast to this. the method of combining auxiliary squares probably was not invented before A.. 1958) II.24 and when this technique for making magic squares was finally revealed by him. For a good summary of the development of the zero. Once again. iii-iv. 23.25 it has no even ones. it appears as a highly developed system for handling both odd and (doubly-) even numbers. History of Mathematics (Dover ed. 876. 180-84. but it is not found in India until the late ninth century.D. p. New York. This quite easy-looking. yet it could not have happened before the discovery of the zero.D. must have been done by the same process. limiting the range still further. but actually very sophisticated. in the course of what must have been at least two centuries of transmission. Moreover. process of combining squares was probably not discovered through some great leap of intellect. with no indication that he knew how to make it. which has a prominent place in the developed system. 69. with a date equivalent to A.26 Thus." which probably occurred around 1100. Furthermore. any precise date for the origin of this technique is impossible to determine. Smith. See D. E.History of Religions religious tradition of the type that was customarily passed down from teacher to disciple without being committed to writing. however brilliant. III (Cambridge. Although the method may seem quite simple once it has been demonstrated. pp. it is scarcely so obvious as to occur-either spontaneously or by reasoning-to anyone. As usual. 26 The earliest certain use of a zero in India is in an inscription at Gwalior. we shall presently see that this method is not likely to have been discovered until after the invention of "the Indian continuous method. the single even-number example published by Yang Hui in 1275 was only an isolated one.

(See Fig. the Indian mathematicians.) After finding this pair of analytic squares.Islamic and Indian Magic Squares The discovery of the "Indian continuous method" for making magic squares was a necessary first step. This was partly because that square had been difficult to construct in the past. 72. 18. and numerous people were experimenting with them. 27.27 Now. Note that the two derived squares have different basic patterns. Datta and A. and simply adding together the equivalent numbers. each containing one of these basic groups. 9. if the original square was composed of the sums of these two groups of numbers. 27 In fact. Also. said square could be reconstituted by starting with two preliminary squares. 3b). the entire square would have reduced to nine sequences of the digits from 1 to 9. and so had presented a considerable challenge. 45. It was also partly due to the fact that a magic square of nine was large enough to be more interestingbecause the phenomena within the square was displayed on a larger scale-yet it was still not so large as to be inconvenient to construct. At any rate. it would have been quite obvious to the discoverer that. The magic square of nine was a favorite one with those who had learned to make magic squares by a continuous method. knew the trick of "casting out nines" (that is. after this discoveryperhaps long after-if someone tried subtracting these reduced numbers from the regular numbers in the equivalent cells of the original square. until the number was reduced to a single digit). N. 1935). Part I (Lahore. if someone took a magic square of nine made by the Hindu continuous method (as shown in Fig. producing a primary square as in Figure 5a. the method of combining two squares to make a third could easily have been happened upon. and applied this "horizontal addition" to every number above 9. like those in the Islamic nations. 282 . in the course of further experiments with it. This would have attracted some curiosity in itself. See B. the Hindus may have invented the method of casting out nines. and they used this in many ways. 36. 63. then continuing to add. 5b. pp. 180-84. taking any large number and adding its digits. Singh. 5b) follows the pattern shown in Figure 4a. History of Hindu Mathematics. The root square (Fig. 54. about the twelfth century. Then. and presumably in India as well. Once this had been found. magic squares of nine became very popular in the Islamic world. which could now easily be surmounted. he would have gotten nine consecutive sets of the root numbers: 0.

However. Still further experimentation would have shown the discoverer of this phenomenon that it was possible to use either of these patterns. in this case. as it moves toward the right. 1 3 1 3 1 0 10 20 5 15 16 8 25 12 4 of FrG. 4 5 2 3 1 2 4 5 3 4 5 2 5 1 2 4 5 15 0 10 20 10 20 5 15 0 15 0 10 20 5 20 5 15 0 10 b. to produce a different distribution in the rows. when they used two preliminary squares in the pattern of the primary square (Pattern B). move in chess. 2 3 4 5 6 7 8 9 a. However. When they used two squares in the pattern of the root square (Pattern A). because it is a secondary pattern.History of Religions with the numbering of each sequence starting one cell higher in each successive column.-Analysis primary square (Pattern B) made by "casting out nines". to be discussed later. 3b): (a) FIG.28 the result was a pandiagonal knight's-move square. 22 14 1 18 10 3 20 7 24 11 9 21 13 5 17 15 2 19 6 23 c. (b) root square (Pattern the numbers in the primary square from those in the A) made by subtracting original square. (A pan2 3 4 5 a. as shown in Figure 6c." although it comes first here. of the type shown in Figure 4c. made by Hindu method 28 I have chosen to call this "Pattern B.-Magic square by the knight's combining preliminary squares. 45 54 63 72 0 9 18 27 36 3 5 54 63 72 0 9 18 27 36 45 63 72 0 9 18 27 36 45 54 72 0 9 18 27 36 45 54 63 0 9 18 27 36 45 54 63 72 9 18 27 36 45 54 63 72 0 18 27 36 45 54 63 72 0 9 27 36 45 54 63 72 0 9 18 36 45 54 63 72 0 9 18 27 3 5 4 5 6 7 6 7 8 9 2 4 4 5 6 1 3 1 of the Hindu continuous magic square of nine (Fig. The Hindus chose to turn over the second (root) square. to make two uniform preliminary squares with the primary numbers and the root numbers arranged in exactly the same order. whereas the primary square (Fig. for a symbolic reason. individually. 5a) has its sequences beginning two cells higher each time." 283 . 6. 5. in order to produce the required sums. more complex than the one which I have called "Pattern A. 4 5 6 7 8 9 1 2 3 6 7 8 92 1 2 3 4 5 8 9 1 1 2 4 3 4 6 7 8 9 1 2 5 6 7 8 9 2 3 4 7 8 9 1 9 1 2 3 5 6 7 8 b. they achieved a lozenge square. one of the pair would have to be flipped before adding.

their diagonal lines intersect like the crossing of threads. to produce a symbolic interweaving of numbers.29 Hence we can understand one reason for the great popularity of this method of construction among the Hindus. It would be deceptively easy to conclude from this distribution of the knight's-move squares that they were actually an Islamic invention. which conceived of the Universe as a woven fabric. 66. they would doubtless have reversed it. repeated in diagonal lines. these two highly symmetrical. when the latter are joined in pairs. the preliminary squares of Pattern B. Thus. so the upper square can be turned over the lower one to produce four different. and this is not only because it can be used to produce pandiagonal squares. Due to this particular design. Actually. is one in which not only the main diagonals produce the constant sum. The preliminary squares of Pattern A have each digit. we have again a clear example of the choice or rejection of a certain type of magic square on purely ideological grounds. have such a broad distribution of the numbers that no single digit appears more than once in the same diagonal (this is what makes the resulting magic squares pandiagonal). yet opposing patterns can be rotated in relation to each other.) The finished square is equivalent to an Indian continuous magic square made with the knight's move in chess. used to make the knight's-move square. but also the broken diagonals. 29 See Kramrisch. Pattern B is perhaps the most interesting and most practical pattern that can be used for making magic squares by the combining method. because they preferred progression to the right. they do not seem to have liked the knight's-move square -none is shown in Narayana's book-although it was very popular among the Indian Muslims. or root number. 284 . This symbolic effect was fully in accord with the basic Hindu idea. then dropped.Islamic and Indian Magic Squares diagonal magic square. when one of these preliminary squares is reversed over the other. Unknown India. and therefore the symbolism of an interwoven universe is totally absent. fall into a very distinctive. However. p. we may recall. but if the Hindus had chosen to use this kind of square. On the other hand. previously alluded to. by means of which is created the magic square proper. If so. and the five 10's in the root square. Notice in Figure 6 that the five 3's in the primary square. when the second square is turned over to combine its numbers with the first. but the idea could have been discovered in India. equally spaced design.

either. 31 See S." Natural History. Thus. why were they so devoted to this method of combining squares which was more complex and time-consuming ? We have already seen that one reason for liking this method was that-at least with preliminary squares of Pattern A-the technique involved a process of interweaving. from unit-centered up to n2 at the center. as illustrated above. and."30) Since the Hindus did not appreciate these squares. and made continuous squares by the knight's move without bothering with preliminary squares. and the single-step diagonally. it does not seem very likely that anyone would have tried to use that somewhat erratic mode of progression (which sometimes entails awkward break-moves) unless he had seen-through an example like the one in Figure 6cthat it was not only possible.31 and even allowing for the fondness for chess among the Indians who had invented the game. one might appropriately speak of Pattern B as the "pinwheel pattern. secondly. In the first place. Cammann. all pandiagonal." because the upper square can be rotated 180? on the lower one to produce a second knight's-move pandiagonal square. there are six possible squares of this type-the two from pattern C. simply took over that mode of progressing. 407-08. the knight's move is simply a logical combination of two other moves in early chess: the single-step straight forward. as a means of constructing magic squares. 32 Someone might consider that the knight's move itself is an erratic type of progress which is not likely to have occurred to anyone unless he had first come upon it through casting out nines on a magic square of nine. 285 . the game of chess must have been invented long before these magic squares. But there are two arguments against this. but highly practical. One might suggest that the men of Islam could have found how to make continuous magic squares by the knight's move without having to see an example made by the method of combining two squares. they apparently did not discover this. not in two colors like ours). which 30 Still another pattern (Pattern C) we might call the "half-pinwheel pattern. LV (1946). once they learned about knight's-move squares. with the four from Pattern B-for any number of the sequence as the middle number. but it seems relatively safe to assume that the first knight's-move magic square was discovered by accident (in India or Persia) when someone happened upon it while experimenting with the Hindu method under discussion. and since the Islamic square-makers.History of Religions knight's-move magic squares. they never seem to have gone on to discover this striking effect. and among the Persians and Arabs who borrowed it from them. Because. "Chess with Mongolian Lamas.32 If the Hindus knew the swift and easy "Indian continuous method" for making magic squares. (Because of this quality. in spite of the resemblance of the magic square's basic grid to an Oriental chessboard (which usually has all its "squares" white.

see T. root square is "dominant" because of the higher values of its numbers and their wider range.33 Siva was a generative god. However. Furthermore. However.Islamic and Indian Magic Squares was symbolically important for squares that were to be used both as religious symbols and as magic charms. 35 chddaka. nor those Chinese examples that were constructed on a Lo-Shu frame (as described in Part I). the union of the preliminary squares was a mathematical expression of mithuna. although even 33 See Dvivedi's introduction to Narayana's G-K (1942 ed. but more especially because it determines the sequences in the finished square.34 N&aryana called "the coverer.38 nor bordered squares.). See P." In view of all this. he named "the covered. symbolized the union of the active and passive principles in the Universe."35 while the more delicate primary square."36 and he spoke of "the coverer" facing down upon "the covered" to produce the magic square proper. One cannot use this system to make the symmetric squares for singly-even numbers. so it was far from successful. masculine character in relation to the other auxiliary square. The dominant or "heavier" square. 37 This was a coincidentia oppositorum. 34 The second. 38 Philippe de la Hire claimed to be able to make magic squares of singly-even numbers by this method. pp. their order as well as the direction of their flow. 1705 (Paris. one occasionally finds an enthusiastic amateur stating that any magic square can be made by this method. that is too extravagant a claim.37 Because so many magic squares. which is numerically weaker and more "passive. the sexual analogy seems rather logical and not as farfetched as it might first appear. samputa. and even different kinds of magic squares. his name for this process." in Memoires de l'Academie Royale des Sciences. in turn. containing the larger multiples. like the Chinese yin-yang symbol but far more graphic. the sexual union of a deity with his consort-which. it has a controlling. According to him. In these terms. which we call the root square. is also a term referring to sexual intercourse. the art of constructing magic squares was revealed to an Indian prince by the god 8iva. Narayana's book provides still another reason. "Sur les Quarr6s magiques. Thus. pp. This is revealed by the language with which he describes the method of construction. For another reference to a connection between the god 8iva and magic squares. containing the lesser sequences. 377-78. 439-40. and Narayana makes it clear that there must have been an idea of symbolic generation in the creation of a magic square through the agency of two complementary auxiliary squares. pp. 1706). can be made by the method of combining the two preliminary patterns. 36 chddya. Katalog der Sanskrit-Handschriften des Universitdts Bibliothek zu Leipzig (Leipzig. But actually he made a square by combining two others and then had to make a great many transpositions before this square was really magic. de la Hire. 1331. 1901). Aufrecht. 286 . No. iii-iv.

The first started by making a natural square (beginning.. then filling these marked cells before returning to fill the rest). Taking the empty square. and when he reached the bottom right corner. p. or by composite methods involving their favorite magic square of fourthis left only those for the singly-even numbers (6.41 The second started rather differently. in a manner recalling the Islamic "double sequence" method for making magic squares of eight (which required first marking the diagonal cells in an empty frame. the Hindu mathematician would mark certain cells.History of Religions some of these aberrant types may be profitably analyzed by breaking them down into primary and root squares in order to investigate their construction. proceeding from right to left in each row as he moved upward to fill 39 Such analysis often reveals striking patterns. 295 ff. it is generally called "the system of broken reversions. rather different methods of procedure within that basic system. they had a special technique. one gets a primary square consisting of nine little Lo-Shu squares. starting at the upper left and counting to the right across each row."40 Actually. For example. See G-K (1942 ed. 14. one finds a primary square composed of a very repetitious Lo-Shu pattern. in which each subsquare contains the same digit nine times. beginning again with the same sequence.39 Since the Hindus had no trouble making magic squares for odd numbers. when one applies it to the Islamic composite magic square of nine (made from the columns of a natural square). 10. when one applies it to the Chinese "Giant Lo-Shu" square of nine (made from the rows of a natural square). 40 For a modem discussion of this system. S. following a definite pattern that usually stressed the main diagonals and a large central block-and often certain other sections as well -keeping a two-part symmetry. Then. the Hindus had three. 1960). but. New York. by C. etc.). for squares of singly-even number above six. reversing certain rows and columns as well-and finally exchanging several pairs of numbers. and since they could easily do those for doubly-even numbers-either by combining special preliminary squares. then reversing the diagonals-and. he filled the marked cells. For these. 287 . followed by a final exchanging of some pairs to break the symmetry. Andrews. at the upper left). p. 386. see W. This involved taking the natural square (made by simply filling an n x n square with the numbers 1 to n2 in regular succession)-or another kind of preliminary square-and reversing the direction of certain groups of numbers. Magic Squares and Cubes (Dover ed. 41 Narayana's third magic square of six was made in this way. he would reverse the process. Planck. as usual.) which have always given problems. Thus.

Because several magic squares of singly-even numbers that were made by the "second Hindu variation" are illustrated in the Persian book in the British Museum.45 In this connection. 7713.. 44 See Kramrisch. retracing his steps to fill the empty cells.44 When the initial interweaving of the numbers has been completed. Unknown India. The former is also shown in Fig. His first magic square of ten and his square of fourteen are shown in ibid. This mode of proceeding is by far the cleverest ever devised for shuffling numbers to achieve a more balanced distribution. which was done more than a century before Narayana wrote his. then back toward the left in the second..Islamic and Indian Magic Squares the cells that he had previously passed over. p. the maker began by counting out the sequence as he moved to the right across the top row. 288 . as he moved upward. 45 See British Museum MS. those in the two central columns and a few in the two center rows (as shown in Fig. 10. it is still necessary to exchange some of the numbers-specifically. 112. was illustrated by Narayana with several examples: squares of 6. we must also consider the strange. When this square is tipped on its right side. p. again changing direction with each row. finally required some last exchanging of numbers.. and 14. 357) was basically made by this technique. 7). The general idea of making magic squares by a method of broken reversions must have been invented long before Narayana's time-perhaps by Islamic scholars. pp. pp. This. only semi-magic.43 After marking a regular pattern in the cells of an empty square. changing his direction with each line. Furthermore. square of ten which was published in China by Yang Hui-a little later. above. Add. 66. p. it embodies a typically Indian mixture of the aesthetic with the metaphysical: the latter being revealed here in the clear analogy to forming the warp and weft of the Universe. When he reached the bottom left corner. but it also involved exchanging halves of the main diagonals. 7. 43 Narayana's first and second magic squares of six (the latter defective) are shown in ibid. as in boustrophedon writing. 384 and 386. so this makes the symbolic correspondence to actual weaving seem even more complete.42 The third variety. he reversed the process. too. in an extension of their method for making magic squares of eight by writing double sequences. But these final steps recall the process of tying off the loose threads on removing a fabric from the loom. it proves to be a primitive form of the third Hindu variation for handling 42 Narayana's second magic square of ten (ibid. a more complicated form of the second. 384 and 385. but still many years before Narayana's book appeared.

36 2 34 3 5 31 FIG. 23. by third method of broken reversions: (a) filling marked cells by boustrophodon numbering. followed by exchanging vertically each pair of adjoining numbers in three columns. I failed to understand the proper construction of this square. above. so the problem of its origin and transmission remains unsolved. and. p. in Fig. in a simple pattern of filling the alternate columns. Yang did not understand this method of construction: a sure sign that the square did not *1 11 24 23 2221 26 a. 21 20 U 12 26 27 10 29 7 C. inverting the third column. 289 . *6 1 35 34 33 32 6 25 11 27 28 8 30 13 14 15 16 17 18 24 23 22 21 20 19 12 26 10 9 29 7 36 2 3 4 5 31 13 14 15 16 17 18 1 35 4 33 32 6 25 11 9 28 8 30 14 (l 16 17 22 323 . 22. I now see that I was completely wrong. Thus. originate with him. On the other hand. after exchanging two pairs at bottom of middle columns. 23. still another step was needed to make the two main diagonals add up properly. "Old Chinese Magic Squares. but it was never done. I presented an analysis of the numbers as they appeared in the tilted square. This was a simple matter of exchanging only four pairs of numbers.History of Religions singly-even squares. However." Fig.47 The fact that he had borrowed for his set a typical Hindu magic square of eight might suggest that this square of ten also had an ultimate Indian origin. 44. 7. 47 See n. disregarding the text. (b) remaining cells filled by proceeding from below.-Hindu magic square of six." and may have even invented the idea. (c) completed square. Although some of the Islamic scholars knew the Hindu methods for handling squares of singly-even numbers by "broken reversions. and exchanging the circled pairs. they seem to have 46 See Cammann. altering direction in each row. 44. believing this to be the probable method of construction. in order to make the rows "magic. 36 *8 20 19 29 '31 b. When writing this previous article." as the columns already were. its numbering begins at the upper right and proceeds to the left in a characteristically Islamic fashion. p.46 Its numbering proceeds in double sequence.

shows bordered squares of eight and ten. which resulted largely from mutual interborrowings-it is still possible. 49. or else one of the bordered squares. At the same time. because one tradition cannot be fully understood without the other. Conversely. used bordered squares. THE ORIENTAL HERITAGE IN EUROPEAN MAGIC SQUARES The earliest numerical magic square in Europe was the magic square of three. to distinguish between certain typical Islamic magic squares and some typically Hindu ones. as we have seen. discussed in the early twelfth century by Abraham ibn Ezra of Toledo. To make magic squares of six. p. 1921). the overlappings make it very difficult to discuss either the Islamic squares or the Hindu ones in isolation. disturbed them as an example of disharmony. in a style that became 48 The Muslims of North India (or modern Pakistan). they customarily still used the well-balanced technique that first appeared in the encyclopedia of the Brothers of Purity. even though some confusion may still remain as to when and where some of the construction methods were actually invented-due to certain overlappings of technique in the Islamic world and in India. Probably the breaking of the symmetry required by the Indian methods.49 He wrote it with Hebrew letter-numbers and interpreted it in a numerological fashion.slam. Buch der Einheit (Sefer ha-Echod). since they both contributed to the development of magic squares in Europe. being under strong influence from Persia. In fact. the Hindus never seem to have taken much-if any-interest in the bordered squares. for example.Islamic and Indian Magic Squares deliberately avoided using them very often. made by them. This further instance of marked cultural preferences provides still another example of the way in which the peoples of certain nations or cultures favored the particular kinds of magic squares that contained elements which corresponded with their most cherished beliefs. trans. 48 probably because there was nothing about them that coincided with their ideology. these two intermeshed traditions share something else in common. and the resulting imbalance. 49 See Abraham ibn Ezra. even though the latter were in use among the North Indian Muslims. which was not appropriate for numbered diagrams that were essentially symbols of an ordered universe under the control of God. 290 . Thus. Ernst Muller (Berlin. The Qanoon-e-. and they always used the highly symmetrical bordered squares for handling singly-even numbers above six.

are frequently discussed in S. who accidentally came across it in the Royal Library in Paris.53 The earliest known European set of magic squares is found in a Latin manuscript of the fifteenth century now in the Jagiellonian Library in Cracow. D. 27. 55 See the comments on Picatrix by A.sen. Economic Foundations. 201. see p. and as far as India-who must have been the intermediaries through whom magic squares were brought to Europe. First we find the magic square of three in an Islamic form. Bing in his foreword to "Picatrix. Poland. by H. but even further developed.. MS 753. See Tannery. 1920). bringing into Europe a new store of Islamic astrological and cosmological lore that in turn helped to build the tradition of the Kabbalah. by a group of Jewish and Christian scholars at the court of Alfonso the Wise. p.50 In fact. a translation into German from the Arabic. it would seem that it must have been scholars in the Jewish communities of Spain. 21 for example. JItude sur les origines et la nature du Zohar (Paris. Plessner. 291 ." das Ziel des Wei. Ritter and M. It seems probable that magic squares may have been not only transmitted. An English translation of this. 898 when general methods for making squares were not yet known. then 50 Ibid. "Le trait6 de Manuel Moschopoulos sur les carr6s magiques.52 his influence is not apparent in the earliest European set of magic squares. IV.54 (Picatrix itself is supposed to have been translated from Arabic sources in thirteenth-century Spain. in 1886. 53 This treatise was first translated by Philippe de la Hire. southern Italy and other Mediterranean lands-which had commercial contacts in the Middle East. if not their actual inventor. XXVII (1962). and S. 1901). i-ii. and his now famous treatise does not seem to have been known in Europe until the early eighteenth century. 173. appears in Scripta Mathematica. but his translation was never published. 51 The Jewish traders of the Mediterranean world. Warburg in Der Islam. 54 Jagiellonian Library. 52 The best study of Moschopoulos and this phase of his work is by Paul Tannery. 15-26. Goitein. "Traite. III (1912).55) The examples in this Cracow set indicate a very early and elementary stage in the knowledge of magic squares. bound together with a fifteenth century manuscript copy of Picatrix. but more exact information about the nature and the extent of this development must await the further study and publication of the numerous books and papers recovered from the Cairo geniza." First written in Paris. this is included in his Memoires scientifiques (Paris and Toulouse." p.51 Although Manuel Moschopoulos has been cited as the introducer of magic squares to Western Europe. who had commercial relations with India and beyond. by John Calvin McCoy.D. VIII (1941). A Mediterranean Society. recalling the first Islamic series of A. I. I am deeply grateful to Dr. and the valuable introductory notes by G. Karpe. 27-60.History of Religions popular with the later Kabbalists. Studies of the Warburg Institute. in the Jewish communities around the shores of the Mediterranean. Zofia Ameisenowa for calling my attention to this and for sending me a microfilm copy of the pages which contain the magic squares.

That may have been a secondary consideration. in Islamic fashion. but the primary reason for including it seems to have been to provide a symbol of Mathematics among the other sciences. by then. 1943). 1603). 59 These squares were published in Ioannes Huser's Zehnter Theil der Schriften Theophrasti Paracelsi (Frankfurt-am-Main. Paracelsus: Sdmtliche Werke (Jena.58 The magic squares of seven and nine in the Cracow manuscript were both done by the Persian continuous method. 493-98. composed by writing the numbers in the primary squares from right to left. this particular square was borrowed some years later. and he also did not realize that the magic square of five could have been done in the same way. but. 131-38. claiming that these could be employed for magic purposes if they were figured on amulets of specific metals associated with the individual planets. I.Islamic and Indian Magic Squares a rather simple magic square of four-though not the simplest one. by the German artist Albrecht Diirer. in 1516. whereas we find here the exact Hindu form. 58 Four pairs of numbers had to be exchanged at the end in order to make this square come out right. Edwin Panofsky.59 56 A great deal has been written about "Diirer's magic square. However. the British Museum's Persian manuscript (Add. Paracelsus: Sdmtliche Werke (Munich and Berlin. The same set later reappeared among the writings ascribed to Paracelsus (Theophrastus Bombastus von Hohenheim). but the excessive number of garbles in the square of nine shows that the writer did not have the slightest idea how the system was supposed to work. Add. 1933).57 and the square of six was made by a rather clumsy handling of the predominantly Hindu method of broken reversions. for inclusion in his famous etching Melencolia 1. assigned to Jupiter in Renaissance Magic. Albrecht Durer (Princeton. He and others have stated that this particular magic square. for example. that is not unique. 7713) shows it "backwards. as we find a somewhat similar method used to make an Islamic magic square of eight in the British Museum MS. Appendix. Obviously he was copying something that he failed to understand. XIV. (Incidentally. which is made by reversing the diagonals in a natural square. pp. 956-61. 91b. IV. which had rather pathetic results. 7713. gen. this Cracow set is notable as being the earliest known instance in Europe of magic squares dedicated to the Seven Planets. they were even more garbled. Theophrast von Hohenheim. 156-71. in the belief that they contained some of the imagined powers of these planets.56) The magic square of five in this set was the common Hindu one." that is. though in different ways (their numbers descend in opposite directions). In spite of all its clumsy errors." See. shown in Figure 4c. p. Part I. and in modem times by Karl Sudhoff (ed. The degree of misunderstanding is further shown by an attempt to apply this method again for making the magic square of eight. 57 Although this magic square of five was already present in the Islamic world in the thirteenth century. Although 292 . was placed in the picture to counteract the influence of Saturn. and by Bernhard Aschner. 1932).).

M. he had not presented a magic square of six.60 Here. and had then been brought from there to Italy by Jewish or Marrano refugees. in a letter. and even that may not have been his own writing. Its link with Kabbalism seems definite. because four years later. before he left for Italy. Universitats Bibliothek MS. Jr. showing that he had not taken the idea from Agrippa's book. although he linked the planets with the squares in reverse order. probably in Spain. whereas his doublyeven squares had begun at the upper left.History of Religions The second stage in European knowledge of magic squares is shown in the far more sophisticated set published by Paracelsus' great contemporary. kindly informed me. the individual squares were once again assigned to the Seven Planets. 1531). the same set was published by Girolamo Cardano. they were certainly borrowed for the MS in which they occur." Journal of the Warburg and Courtauld Institutes. with specific instructions telling how to engrave them on amulets for working good or evil. 61 Charles G. author of Agrippa and the Crisis of Renaissance Thought (Urbana. because the same these squares have been ascribed to Paracelsus. in 1539. chap. Agrippa published the set in both Arabic numbers and Hebrew letter-numbers. Perhaps the most absurd and inaccurate treatment of them appears in K. seven. Nauert. since they did not appear in the earlier manuscript copies of his book which he sent out in 1510. Although the squares of five..50). xxii. item 39 (no pagination). Nowotny. His "explanations" are shockingly bad because he failed to understand the workings of magic squares himself. 293 . Cornelius Agrippa von Nettesheim. and eight were all of more direct Islamic origin. 62 Girolamo Cardano. Many foolish statements have been made about Agrippa's set of squares. Practica arithmetice et mensurandi singularis (Milan. chap. This appears in his famous book De occulta philosophia. who also associated them with the Seven Planets. De occulta philosophia (Cologne. Probably he had learned these squares during his residence in Italy (1511-18). we might assume that this group of squares was circulating in early sixteenth-century Italy. 46-57. once again the transmission does not appear to have come through him. xlii. where he had had contact with Kabbalistic circles. 1965). and. 1539).ch. in 1531. and nine are all made by the Persian continuous method which Moschopoulos had described. 60 Cornelius Agrippa von Nettesheim.62 It seems probable that this second collection of magic squares had been assembled in Hebrew Kabbalistic circles.9. For the squares of four. XII (1949). in a chapter devoted to planetary lore. forced to leave Spain after the Expulsion of 1492. six.61 Furthermore. with their numbers beginning at the upper right in Islamic (or Hebrew) fashion. that there are no magic squares in the manuscript of the earlier version now in Wiirzburg (Wiirzburg. "The Construction of Certain Seals and Characters in the Work of Agrippa of Nettesheim. A.

This new approach toward them actually began in sixteenth-century Germany. However. these essays into Numerology were quite arbitrary and rather farfetched. pp.Islamic and Indian Magic Squares set reappeared once again in purely Jewish form in the Kabbala denudata. 66 Although the custom of making and carrying metal talismans inscribed with magic squares lasted into the eighteenth century (see n. 65 Athanasius Kircher. to avoid profaning the Divine name. Arithmologia (Rome. 1665). as problems in mathematical arrangement. to try to find Christian messages in the numbers of these squares. The superstitious employment of magic squares in Renaissance Magic. 67 Adam Riese published several mathematical textbooks with slightly varying names.64 Baron von Rosenroth carried still further some previous efforts. the squares were only considered most rationally. with Michael Stifel and Adam Riese. marked the final step in the death of their function as significant symbols-although that function had probably already been lost in the course of their transmission to Europe. and the equally stupid attempts to reinterpret them in terms of Numerology. above). dating from the sixteenth to the eighteenth centuries. The earliest editions lacked the magic squares. with the help of Jewish friends. X. while Adam Riese demonstrated how to use the Persian continuous method to make oddnumber squares likes those that had been published by Agrippa. 477.. 64. and apparently this prevented these mathematical books from reaching France. Renaissance preoccupation with magic squares as supposed vehicles of occult because the Europeans powers began to die considerably before that-probably found that these failed to accomplish the many wonders that were claimed for them. in 1684. pp. p. so they do not seem to have attracted much attention outside of Christian Cabalistic groups.66 Henceforth. the earlier. but these appeared in the 1550 edition of his book Rechenung nach der lenge (sic). S. many of which have numerous metal pieces bearing magic squares. compiled by Baron Knorr von Rosenroth. thus.63 Not only were the squares here printed entirely in Hebrew letter-numbers. in the increasingly more materialistic Western World. 64-109. 103a-105b.67 The wars and social upheavals that attended the Reformation in the various states of Germany tended to isolate them from the rest of Europe. and 150-69. thus giving half of the name of God (YHWH). by Athanasius Kircher65 and others. as we have previously noted. but they also had the Jewish convention of writing 15 with the numbers 9 and 6. 294 .J. This avoidance makes it possible to distinguish Jewish talismans from pseudo-Jewish amulets in European coin collections. 64 The number 15 would be written in Hebrew letters with yod and he. which was the next area to develop a 63 The life and work of Baron Knorr von Rosenroth are discussed in The Jewish Encyclopaedia. The former described the bordered squares. and. to draw back into the Church people who had become interested in such "occult" matters.

When he later published a two-volume account of his travels (in 1691).69 In this.P. One of them. not suitable for making very large squares. did not contain any magic squares.68 Actually. 295 . and in an English translation in 1693. rather primitive method. beginning in the early seventeenth his book had been century. in Amsterdam again in 1713. Then he demonstrated the faster way of making the odd-number squares that had been published by Agrippa. Was this because the connection of "Agrippa's squares" with the Kabbalistic tradition.) 69 Simon de la Loubbre. Du Royaume de Siam. the Europeans tended to ignore this second type-the Persian continuous method-and from that time on. showing that they. he described at some length the newly learned Hindu continuous method.History of Religions keen interest in magic squares. he rendered it uncomprehendingly and far too literally. happened upon the "Diamond method" for constructing magic squares (described and illustrated in Part I). 1624). Nevertheless. of the French editions. was whiling away his time on shipboard by constructing magic squares with "the Method of Bachet de Meziriac. the French mathematicians were chiefly preoccupied with seeking some general methods for constructing magic squares. Europeans have known it by his name. and he spent the rest of the voyage experimenting with it. too. he devoted a whole chapter to the subject of magic squares. were properly done by an analagous system of continuous numbering. 235-88. Simon de La Loubere was delighted with the ease and speed of this technique. published in 1612. while investigating how Agrippa's odd-number squares had been made. Problemes plaisants et delectables (Lyons. returning from a diplomatic mission to Siam. but it was hailed as an exciting new discovery. ever since. (The earlier edition of this book. Simon de La Loubere. In 1688. a French physician named Vincent." is poor. offered to show him a faster way." which the latter had learned on a visit to Surat. they devoted all their attention to the Indian one. The English translation by "A. published in French in Paris and Amsterdam in 1691. Bachet de Meziriac. rather than by the clumsy "diamond method" rediscovered by Bachet de Meziriac. This was the "Hindu continuous method. The chapter on magic squares is found in II. G. now in dis68 Claude-Gaspard Bachet de Meziriac. this is a slow. and. and in a book published in 1624 he advocated this as a practical means for making any odd-number magic square." when a fellow-passenger. C. Having Agrippa's squares-since widely circulated before the troubles set in-but lacking any literature on how to make them. and its section on the magic squares is worthless. Because the translator did not understand the subject.

A copy of this book was sent to the Royal Academy of Sciences in Paris. Traite des Quarres sublimes (Brussels. taught in Europe at that time as a fundamental aspect of mathematical training. above) only recalled to him a method he had thought of himself. during his shipboard experiments with the Hindu continuous method. Vincent.Islamic and Indian Magic Squares repute. 1706). in 1704. . three for 15. or of root numbers. ("Casting out nines" was another facet of the legacy from the Islamic world. he showed a number of squares consisting entirely of primary numbers. for the discontinuity between the 1 and the 2 (as shown in Fig. derived from analyzing other magic squares..74 72 Abb6 Poignard. 73 See the Histoire de l'Academie Royale des Sciences. one for 5. of Math. published a book of his own about them. 72. to which he joined symbols of the multiples of the root number (5) by adding small marks beside the primary numbers (none for 0. knew La Loubbre and the latter's work on magic squares." In the course of an analysis of the Hindu continuous method. pp. In any case. p. M. Hist. after its publication in Paris and Amsterdam in 1691. 1704). two for 10. 70 Ibid.70 Perhaps he had gotten the idea from Dr. 3b) kept the arrangement of the numbers from being so immediately obvious. Therefore. he drew a primary square like that in Figure 6a. Du Royaume de Siam.72 In this. which passed it on for review to Philippe de la Hire. but this sounds rather lame. too. because of its value for checking the results of other operations. 152-54. For a clearer explanation of this. 72. Grand Canon of Brussels (midway between Amsterdam and Paris). this little diagram probably served as the initial inspiration for the European "invention" of the Hindu method of combining auxiliary squares. it was "more of a trick.. immediately became very popular. and thus the uninitiated could not see as readily how it was prepared. p. de La Loubere tried to describe its workings by discussing the use of primary numbers and root numbers. or possibly he had just happened upon it by "casting out nines" on a magic square of nine. 267. 1705 (Paris. La Loubere's book. 296 71 See Smith. who became an enthusiastic student of magic squares and. 74 La Hire claimed that Poignard's book (n. had discredited them ? More likely it was because although the Hindu continuous method was not any easier-nor more difficult-it seemed more effective as a puzzle.73 and the latter was inspired by it to publish an account of the method of making magic squares by combining two preliminary squares. to be able to reproduce it.71) At any rate. Among its readers was apparently the Abb6 Poignard. he. while the latter was teaching him what he had learned in India. and four for 20).

76 though he did not say whether he had gotten this. M. Vincent. One of the pioneers in working with these was Bernard Frenicle de Bessy. he demonstrated. had so absorbed the scholars of Islam. Sauveur himself gained a high reputation for his experiments with magic squares and his deep knowledge of the field. who is often given the credit for devising them.79 In this. once again. d'Ons-le-Bray. and on the topic of concentric bordered squares. He illustrated the latter by a splendid magic square of ten. p. "Construction g6n6rale des quarr6s magiques. too. 285. who first demonstrated that there were exactly 880 magic squares of four-thus finally solving the problem which. he invented magic cubes. 77 In the Memoires de l'Academie p. 1711). 79 Loc. being frequently cited later. the "new" technique became widely known as the "Method of La Hire.) 78 See reference in n. or 246 of the one-volume London edition. M. Sauveur. nor the Hindus before them. and it aroused much interest..77 while further investigation of the latter was carried on by both La Hire and Sauveur. (The earlier volumes of the Mimoires were given numbers." and neither Sauveur. 92-138. 1691). 75 above.75 who also published his findings in a memoire for the Royal Academy. and for making singly-even squares on cores consisting of one or more fourth-order squares within 75 M. as well as a clever method for making the singly-even ones. the old Islamic techniques for making magic squares of the higher doubly-even numbers by combining in one pattern a series of squares of the fourth order. received any credit.History of Religions This appeared in the Mdmoires of the Royal Academy for 1705. nor La Loubere. After Frenicle's death. his method was improved upon by Joseph Sauveur. in spite of this. other French scholars had also become interested in the doubly-even magic squares. pp. 76 Du Royaume de Siam. nor Poignard. and he developed the idea of lettered squares (the so-called Latin squares). Meanwhile. V (Paris.78 which were later taken up by the Swiss mathematician Euler. instead of dates. 297 . Five years after this. Royale des Sciences. but. de La Loubere finally described a way to make magic squares for any doubly-even number (the old Islamic one. II. from Dr. of marking the diagonal cells and using double sequences). 1710 (Paris. cit. Philippe de la Hire published his writings on these. Among other things. made with the second Indian method of broken reversions." in Memoires de I'Academie Royale des Sciences. several centuries before. 241 ff. The final distinct contribution to the construction of magic squares in France was a memoire for the Royal Academy written in 1750 by M.

he had much to say about "Nasik squares. and went on to invent a few new squares of his own. The one for Encyclopaedia Britannica was still present in the 11th helped to build and maintain-in Britain and America-an interest in these that carried on into the twentieth century. and his writings for mathematical journals. So. A.) However. Among these was our American scholar. he wrote a number of learned articles about these pandiagonal squares. these very old ideas seemed new. at least. When he returned to England. Benjamin Franklin. XIII. pp. Frost learned the technique of making pandiagonal magic squares of four. We have already noted that Narayana mentioned the Hindu's association of magic dedisquares with Siva. Magic Squares and Cubes. In the latter.Islamic and Indian Magic Squares cleverly contrived bordering frames. the Rev. Benjamin Franklin's ingenious squares were first published in London. through an English missionary. magic squares had finally descended to the status of a mere recreation. H. who spent some years in the old city of Nasik. sacred to Siva. they continued to interest a great many people. which was probably carried over to India.80 The last direct influence from Asia in the field of magic squares reached Europe in the middle of the nineteenth century. Mr. "The Franklin Squares. (Unfortunately. 310-13. and the holy city of Nasik-especially cated to his worship-was still noted as a center for their study. as such." or simply "Nasiks. as well as his pioneer article on magic squares for Encyclopaedia Britannica. in 1769. 81 One of A. Frost. While there. from the twelfth to the sixteenth centuries. H.) The author introduced his demonstrations by explaining that his work on magic squares was the result of his efforts to overcome the boredom of life in the country. 89-112. as well as for its production of charms and amulets on which they were figured. who studied the Memoires of the Royal Academy relating to this subject." Obviously he did not suspect that this type of magic square may have originated in the Islamic world. Frost was effective in communicating his enthusiasm to others." and even used the unfortunate adjective "Nasical. Andrews. which he called "Nasik squares." in W. his awkward and misleading terms still persist among English and American writers. in eighteenth-century France. In spite of Frost's dedicated efforts to raise the study of magic 80 See Paul Carus. but. (In Europe. nor did he know about the long tradition of extensive work with them on the part of the Islamic scholars. and became much intrigued with them. S. Frost's articles appeared in the Cambridge Mathematical Journal for 1857." 298 .

had all long since been completely forgotten. or for simple recreation. of course. And. 299 .History of Religions squares to a higher plane of mathematical study. any idea of the ultimate origin of the European magic squares (except for the obviously-named "Nasik" ones). and even the strange and wonderful properties that made it possible for men to find and hold these ideas. however. or of the metaphysical ideas that had surrounded them in the Islamic world and in India. most people who worked with them did so as a form of hobby.

Sign up to vote on this title
UsefulNot useful