Sie sind auf Seite 1von 10
Nexus Conference 2006; Relationships between Architecture and Mathematics Genova, Italy June 7 ~ 9, 2006 ‘Systems of Monads’ as Design Principle in the Hagia Sophia: Neo-Platonic Mathematics in the Architecture of Late Antiquity! HELGE SveNsHON* Ropotr H. W. Sriche.* Introduction The Hagia Sophia in Istanbul, built between 532 and 537 in the time of Emperor Justinian, is universally acknowledged as one of the few examples of outstanding architecture in the world (see esp. [Mainstone 1988]) (fig. 1). An ongoing stream of descriptions continues in trying to explain this fascinating “wonder of space”. Exuberant praise of the building is to be found as carly as Procopius (Aedificia I 1, 29), the historian of Justinian, who especially and exceptionally highlighted the extraordinary “splendour and harmony in the measures” of the temple. Despite various and extensive scientific discussion the design of Hagia Sophia still escapes complete and sufficient understanding. Anthemius of Tralleis and Isidorus of Milet, the two architects of the Great Church, were among the best mathematicians and engineers of their time. Thus it seems natural to search for latent mathematics hidden in the extent of the building, to determine the substantial structures of mathematical relevance and finally to develop an explanatory model for the design and for the meaning of this spectacular architecture. According to general consensus, one has to start from a barely visible square, which can be found in the floor plan between the four large main piers. With a side length of almost exactly 31m, it is obviously planned with a high degree of accuracy. Such a distance can usually —and at first sight quite convincingly —be translated to exactly 100 ft. Following this hypothesis the diagonal between the piers results in an irrational number of 100 x V2 (= 1.4142135...) and consequently leads to a large number of further irrational building dimensions — caused by their geometrical dependence on the basic square ~ with which it would have been hardly possible to plan, draw and mark out this building precisely (fig. 2). Birkhauser/Springer Verlag 2006 L.This research was sponsered by Zentrum fiir interdisziplinaere Technikforschung (ZIT) of the Darmstadt University of Technology. See preliminary reports: [Stichel and Svenshon 2002; Svenshon 2003; Stichel 2003]. * Technische Universitit Darmstadt, Fachbereich Architektur, El-Lissitzky-Str. 1, 64287 Darmstadt - Germany, svenshon@gta.tu-darmstadt.de, stichel@klarch.tu-darmstadt.de Nexus VI - Architecture and Mathematics 117 Helge Svenshon, Rudolf H. W. Stichel figure 1 Fig. 1 Hagia Sophia, interior (Kahler 1967, PI. 24) Fig. 2 Hagia Sophia, floor plan with the basic square. (Svenshon after van Nice) Fig. 3. System of side and diagonal _ numbers (Svenshon after Heller 1965, 334) figure 3 112 Nexus VI - Architecture and Mathematics ‘Systems of Monads’ as Design Principle in the Hagia Sophia However, the problem can be solved by a surprisingly simple process employing a mathematical ‘trick’. If one assumes the length of the square side to be 99 ft. instead of 100, the diagonal can be calculated at almost precisely 140 ft.; the margin of error is now only within a few thousandths of a foot and thus remains irrelevant for the building procedure (fig. 2). A proposition for this idea can be found in early Greek mathematics: a series of so-called side and diagonal numbers. Theon of Smyrna, in his mathematical explanations of Plato’s works (Expositio rerum mathematicarum (Hiller) 43, 5-8) was the first to formulate it as a rule. Some 300 years later Proclus Diadochus’s commentary on Plato’s Republic (in Platonis rem publicam (Kroll) 11, 248) summarised it within a precise mathematical formula: ‘The Pythagoreans and Plato thought to say that, the side being expressible, the diagonal is not absolutely expressible, but, in the squares whose sides they are, [the square of the diagonal] is either less by a unit or more by a unit than the double ratio which the diagonal ought to make: more, as for instance is 9 than 4, less as for instance 49 than 25. The Pythagoreans put forward the following kind of elegant theorem of this, about the diagonals and sides, that when the diagonal receives the side of which it is diagonal it becomes a side, while the side, added to itself and receiving in addition its own diagonal, becomes a diagonal. And this is demonstrated by lines through the things in the second [book] of Elements by him (Euclid] [Fowler 1990, 101] (fig. 3). Starting with a square with a side of 1 unit (= monas) and according to the rule 8, 8,=8,+d,, 8,=8,td,,... and dy d, a series of squares is developed. Its ratio between side and diagonal delivers at every further step of that series a more and more precise approximation to V2, because the difference between the square of the diagonals and the double square of the sides alternates solely between +1 and~—I, ic., in every case exactly one unit [Heller 1956, 4}. The unit, however, does not by itself determine the elegant result of this rule, but, “similar to a sperm, that carries within itself all the attributes of the future life seminal, is by itself capable to produce this ratio between diagonal and side in a square” [Heller 1965, 335f]. Theon of Smyrna justifies this with unmistakable clearness: “Therefore since the unit, according to the supreme generative principle, is the starting-point of all the figures, so also in the unit will be found the ratio of the diameter to the side.” It is those properties “which give harmony to the figures” [Thomas 1980, 133¢f,]. Nexus VI - Architecture and Mathematics 113 Helge Svenshon, Rudolf H.W. Stichel Thus it is not surprising that the simple practicality of the calculation, combined with the hinted cosmological claim, promotes the side and diagonal numbers an important component of Greek applied mathematics and geometry (/ogistike and geodaisia). Especially in the writings of Heron of Alexandria, under whose name mathematical and engineering handbooks were edited between the first centuries A.D. and the Byzantine middle ages [Cantor 1907, I 381; Meissner 1999, 140 £.], numerous examples can be found that document the application of this method of calculation for handling integer approximations of V2 [Smily 1944, 18-26]. Depending on the needs and the demand for accuracy these values were varied but never deviated from the side and diagonal numerical series and their derivatives. Consequently Didymus of Alexandria uses for the surface calculations of a square block of wood with the side measurements of 10 ft. a diagonal measuring “approximately 14 1/7” (99/7), which corresponds with the decuple of the very accurate approximation of V2 provided in the sixth square of our series [Bruins 1964, 1 125, II 84, TIT 178]. In contrast, Heron in his Metrika (Heron Alexandrinus opera (Schéne) III, 57f.), an extensive and probably original treatise about the methodical visualisation of surface and volume calculations, utilizes the slightly more inaccurate values of the third and fourth square, the 7:5 and the 17:12 ratio respectively, the latter a more familiar approximation for the classical antiquity and the middle ages. Those also appear as exemplifications in the formulas of Theon and Proclus. But the practical advantages of this calculation method were known outside Greek mathematics as well: In the cuneiform texts of ancient Babylon and especially in the so-called coefficient list of Susa, which was written in the early second millennium B.C., the ratio 17:12 is used as aV2 constant (in the Babylonian sexagesimal system in the form of 1:25 = 60/60 + 25/60 = 85/60 = 17/12) [Bruins and Rutten 1961, 25ff. pl.4/5]. Derivations that can only be imaginable with knowledge of the square series, can be found on the ancient Babylonian cuneiform tablets Plimpton 322 [Neugebauer/Sachs 1945, 38ff., 130; Robson 2001, 167-206] and VAT 8512 [Neugebauer 1935, 341 f., pl. 27, 52; Horup 2002, 234]. Moreover, Lennart Berggren’s synopsis of ancient and medieval approximations of irrational numbers [2002, 31-44] shows that the use of single values from the side and diagonal numerical series is verified in nearly all antique cultures. This exceptional phenomenon is easily understood, as approximations for V2 can be demonstrated geometrically in a 114 ‘Nexus VI - Architecture and Mathematics ‘Systems of Monads’ as Design Principle in the Hagia Sophia simple fashion. Theon’s remark that the side and diagonal numbers in particular give “harmony to the figures” has also a strong pragmatic connotation. The exact numerical proportion, which can be calculated with the previously mentioned algorithm, result in the side measurement of a regular octagon, after the choice of the appropriate initial dimensions (s/d numbers). The evidence for this is again supplied by Heron’s Metrika (Heron Alexandrinus opera (Schéne) III. 57f). For the area calculation of a regular octagon with a side length of 10, two right-angled triangles with the rational sides of 5 and 12 and the hypotenuse of 13 are constructed (fig. 4). Next, Heron divides the longer side, which is also the perpendicular bisector of the octagon side, in two sections measuring 5 and 7, whereby an isosceles, orthogonal triangle with catheti measuring 5 is formed over the half side of the octagon. The hypotenuse corresponds approximately to the larger section of the first triangle’s longer cathetus and is logically numeralised with the rational value of 7 in order to accomplish the following calculations of the octagon’s surface area on the solid basis of the side and diagonal numbers. The efforts of the ancient geodesists and logisticians to be able to perform their computations on the foundations provided by expressible numbers, interlinked through a super-ordinate system, hardly become clearer than in this example. The fact that these are not isolated cases of special knowledge, which only a few experts would be able to draw upon, but rather are broadly received common knowledge is proven by a letter of Gregorius of Nyssa written in late fourth century A.D. It was his request for administrative assistance to Amphilochius of Iconium. for the building of an octagonal church [Teske 1997, 85-88]. Besides the detailed description of the structure, precise measurements are mentioned in this letter, which were used in order to calculate the costs for the building materials as exactly as possible. The particular data provided by Gregorius allow the reconstruction of a four- winged structure with an octagonal nucleus, whose dimension between axes (octagon side = 10 cubits, circumcircle radius = 13 cubits, internal radius = 12 cubits) correspond exactly to the ones found in Heron’s Metrika (fig. 5). it is known that the bishop’s friend and colleague, Gregorius of Nazianz, had included Heron among those he considered the three key figures in Greek mathematics [Cantor 1875, 12], along with Ptolemacus and Euclid. Therefore it is only natural to assume that Gregorius of Nyssa was acquainted with Heron’s handbooks. This source illustrates how interwoven the building processes and mathematical thought of late antiquity were. It also demonstrates how the theoretical and philosophical fundamentals figure 4 Fig. 4 Heron’s octagon construction (Heron Alexandrinus opera (Schéne) IH, 57) Fig. 5 The octagonal church according to Gregorius of Nyssa (Restle 1979, Pl. 58) Nexus VI - Architecture and Mathematics 115 Helge Svenshon, Rudolf H. W. Stichel figure 7 Fig. 6 Hagia Sophia, construction of the conchs (Svenshon) Fig. 7. Heron’s circle measurement (Sullivan 2000, PI. 36) Fig. 8 Hagia Sophia, floor plan with the most important measurements (Svenshon) figure 8 ; | of scientific mathematics were introduced to architecture through the application-friendly disciplines of geodaisia and logistike. Anthemius and Isidorus, the architects of Hagia Sophia, were mathematicians and engineers, trained theoretically as well as practically. It is a fact that both of them re-edited older treatises but also wrote scientific handbooks of their own. The imperial request to plan the Great Church put them before a task for which no adequate experience existed. Therefore it is even more probable that for the conceptual work they resorted to well-tried basic principles from ancient handbooks of mathematics, which at the same time they possibly also amended. If onc also considers the central role that the described octagon with its rational side measurements plays in ancient surveying [Cantor 1907, I 559], it encourages the assumption that this figure was used for the building in a greater scale as well. Constructing an octagon from two congruent squares with the side and diagonal lengths of 99 ft. and of 140 ft. respectively by rotating one to the other by 45° results in the division of the squares’ sides in nearly rational segments measuring 29 ft. / 41 ft. / 29 ft. (= 99 ft.), which correspond to the side and diagonal numerical 116 Nexus VI - Architecture and Mathematics ‘Systems of Monads’ as Design Principle in the Hagia Sophia proportion of the fifth square. Precisely those numbers are now used constitutively for further construction of the church’s floor plan. The concept of the concha circles can serve as an example, because they can be derived from the basic geometric figure in a few steps. If one completes the lines resulting at the north-eastem main square sides to further squares with the side measurements of 29 ft. and 41 ft. the intersection point of the two squares also creates the centre of the concha circles that thus obtain their final size with an exact radius of 20 1/2 ft. (= 41/2 ft.) (fig. 6). Following this pattern, the orthogonal structure of the floor plan can be developed simply and using only rational measurements, which nearly without exception follow or descend from the side and diagonal numerical series. Given, however, that the Hagia Sophia is a centralized structure dominated by a dome, a new aspect of calculating with irrational numbers is introduced. In a similar way as with the square side and diagonal, the diameter of the circle also stands in an irrational ratio to the circumference, defined by the number 6 (=3.14159...). However, for the planning of the dome, expressible numbers were essential to make the execution of both exact and conveyable calculations possible, directly at the building site. Greek mathematics offers a practical formula for this purpose as well, which is compatible with the side and diagonal numbers. Archimedes positions the approximation for 6 between 3 1/7 and 3 10/71. Even though this value is very accurate, it is logistically very inconvenient. Therefore Heron suggested the use of the less exact value of 3 1/7 (22/7), which is still completely feasible for all circle-related calculations (Heron Alexandrinus opera (Schone) III, 67). In the many exemplary calculations that follow, the fact that the radius and diameter numbers are only slightly varied stands out. The number 7 comes into use most frequently, followed by divisors or multiples of this value. This is because the number 7 in the denominator can be easily reduced in the circumference formula U=2x6xrand thus produce an integer value for the circumference. Inhis Geodaesia, Heron of Byzantium used exactly those examples for the measurement of circles [Sullivan 2000, 1308] (fig. 7): 70 x 22/7 = 220 (= 140/2 x 22/7) 210 x 22/7 = 660 (= 2 x 105 x 22) These values can be rediscovered in the dimensions of Hagia Sophia. The diagonal of the main square measures exactly 140 fi and defines the circle on which the great pendentives rest. The diameter of the dome is 105 ft. (3/4 x 140) and so further divisors can be derived that are also compatible with the system of the side and diagonal numbers. As shown in the plan in figure 8, the principal architectural measurements of Justinian’s Hagia Sophia may be expressed by Nexus VI - Architecture and Mathematics 117 Helge Svenshon, Rudolf H. Wi. Stichel integer numbers. They all are related by well-defined arithmetical and (also) geometrical proportions. Therefore one might speak of a “system of monads”, though perhaps not in exactly the same sense as this phrase is used in ancient mathematical philosophy, e.g., in Domninus of Larissa in the fifth century A.D. [Boissanade 1832, IV 413] or even earlier in lamblichus (introd. mathem. (Pistelli — Klein) 10,11) in the fourth century A.D., who is allegedly citing Thales of Miletus from the sixth century B.C. [not cited in Diels and Kranz 1951/1952, 1 § 11]. This “system of monads” found in the geometry of the Hagia Sophia was well known in Classical Antiquity and widely discussed not only by mathematicians, but also in philosophical circles, as the creation of the world was understood as a geometrical problem. Asa locus classicus one may cite Plato’s Timaeus: ...all ... kinds were without proportion and number ... in such a condition as we should expect for anything when deity is absent from it. Such being their nature at the time when the ordering of the universe was taken in hand, the god then began by giving them a distinct configuration by means of shapes and numbers ... with the greatest possible perfection [53 a/b, transl. F. M. Crawford (1959)]. Ina paragraph following this quote, Plato declares the isosceles right-angled triangle the very first of all shapes used by the demiurge. When the neo-Platonic philosophers Theon of Smyrna, second century A.D. (Expositio rerum mathematicarum (Hiller) 43, 5-8) and Proclus Diadochus, late fifth century A.D. (in Platonis rem publicam (Kroll) Il, 24f), discuss at length the mathematical phenomenon of the numbers of the sides and the diagonal of the square, they are giving no more than an explanation of Plato’s theories. In the Theologumena Arithmeticae of Pseudo-Iamblichus, awork of ca. 300 A.D., the Diametrikd are again cited in connection with the creation of the world —as in Theon’s work combined with other groups of integer numbers — through which all is determined spermatikos (through the sperm) (Theologumena arithmeticae (de Falco — Klein), 79). In the fourth chapter of Zntroduction to Arithmetic by Nicomachus of Gerasa, a philosopher of Roman imperial times, ‘one might find the same perception, but expressed with a higher degree of verbal complexity: .. arithmetic existed before all the others in the mind of the creating God like some universal and exemplary plan, relying upon which as a design and archetypal example the creator of the universe sets in order his material creations and makes them attain to their proper ends ... (Jntroductio mathematica, transl. D’Ooge 813). 118 Nexus VI- Architecture and Mathematics