Serious investigation of the origin and nature of the Rotas-Sator square began in 1881 with the publication of Köhler’s historical survey in Zeitschrift für Ethnologie.1 A vigorous and protracted debate has been illuminated from time to time by the brilliant researches of modem archaeology; but despite almost eighty years of academic controversy no conclusive solution has yet been found to the mystery of the ‘magic square’.
This construction was a cryptic rebus which appears in two different forms, an earlier and a later. Both consist of a symmetrical combination of five words, each of five letters, the whole forming a square which can be read in four different directions.
ROTAS SATOR
OPERA AREPO
TENET TENET
AREPO OPERA
SATOR ROTAS
The formula has a long history. The earliest text was at one time thought to be a Copic papyrus of the fourth or fifth century A.D.,2 but recent discoveries have now dated it in the Roman period. During the campaign of 1931-2, excavations at Dura-Europus on the Euphrates, conducted under Rostovtzeff by Yale University and the French Academy of Inscriptions and Letters, unearthed three specimens on the walls of a military office in what had originally been the temple of Azzanathkona.3 The following year a fourth was discovered, all of which must have been inscribed before the Persians destroyed Dura soon after A.D. 256. The finds at Dura vindicated a third or fourth century British specimen scratched on a fragment of wall plaster from Victoria Road, Cirencester, in Gloucestershire. Haverfield had long before attributed this to a Roman date,4 but his theory was discounted at the time, since no other instances were then known which could be dated before the Early Middle Ages, and because the only evidence at Cirencester was the letter forms (principally the A’s) and the general Romano-British character of the find spot. Five years after the discoveries at Dura, Della Corte, supervising excavations at Pompeii, came across a version written on a column near the amphitheatre.5 This discovery now led to the proper restoration of a similar, though fragmentary, example he had already published in 1929 from the house of a Publius Paquius Proculus, also at Pompeii.6 The latest specimen to be recovered, a third century Rotas Square from Altofen, Budapest, was found in 1954 and published with a commentary in German and Russian by Szilâgyi in Acta Antiqua Academiæ Scientiarum Hungari¬cæ.7 The formula here appears on a roof tile from the villa publica of ancient Aquincum, the residence of the imperial governor of Pannonia Inferior.
The later history of the charm spans the sixth century to the nine¬teenth and embraces the continents of Europe, Africa, and America.8 In France the earliest example occurs in a Carolingian Bible of 822, originally the property of the monastery of Saint-Germain-des-Près. In the twelfth century it is inscribed on the masonry of the Church of St. Laurent near Ardèche and in the keep of Loches, while in the thirteenth century parchment of Aurillac it apparently intercedes for women in labour. By the fifteenth century it has become a touchstone against fire in the Chateaux of Chinon and of Jarnac and in the court¬house of Valbonnais; but it is not until the sixteenth century that its efficacy as a cure for insanity and for fever is described in two early books, De Varia Quercus Historia, by Jean du Choul (Lyons 1555), and De Rerum Varietate, by Jérôme Cardan, a medical astrologer, (Milan 1557).
Perhaps the most extraordinary case related here is that of a citizen of Lyons who recovered from insanity after eating three crusts of bread, each inscribed with the magic square. This repast was punctuated by the recitation of five paternosters in remembrance of the five wounds of Christ and of the five nails of the Cross: Pro quinque vulneribus Christi, quae moriendo accepit, nee non pro clavibus. This local association of the square with the Lord’s Prayer and the nails may go back to the second century bishop of Lyons, St. Irenaeus, who himself had a devotion to the five ‘summits’ of the Cross: et ipse habitus crucis fines et summitates habet quinque, duas in longitudine et duas in latitudine et unam in medio in quo requiescit qui clavis affigitur.9
Knowledge of the charm was not confined to Europe. In his Arithmologia (Rome 1665) R. P. Kircher relates that on a voyage to Abyssinia he had discovered that the Ethiopians invoke their Saviour by enumerating the five nails of the Cross, namely: SADOR, ALADOR, DANET, ADERA, RODAS – clearly the five words of the square in a corrupt form. A similar usage appears in a version from a tomb near Faras in Nubia where the five words follow a Coptic phrase which has been interpreted to mean “the names of the nails of Christ’s Cross.”10 In the eleventh century, on the other hand, the five words were used in Abyssinia to denote the five wounds of Christ.11
Other regions found other applications for the formula. In Cappadocia, in the time of Constantine VII, Porphyrogenitus (913-959), the shepherds of the Nativity story are called SATOR, AREPON, and TENETON,12 while a Byzantine bible of an earlier period conjures out of the square the baptismal names of the three Magi, ATOR, SATOR, and PERATORAS.13 By the end of the Middle Ages its prophylactic magic was firmly established in the superstition of Italy, Serbia Germany, Iceland, and even North America. The most recent example comes from nineteenth century South America, where it was still in use to cure dog-bites and snake-bites.14
The traditional popularity and astonishing versatility of this charm have led scholars to believe that the words conceal a meaning other than the obvious: “The Sower AREPO (whatever that may mean) holds the wheels with care.” Such convictions have reaped the scorn and ridicule of sceptics who hold that the sole, intrinsic merit of the formula is that it is a perfect palindrome. The composition of palindromes was, in fact, a pastime of Roman landed gentry. Sidonius,15 writing to Burgundius, neatly defines such lines which can be read equally well from end to beginning as from beginning to end: Hi nimirum sunt recurrentes qui metro stante neque litteris loco motis ut ab exordio ad terminum sic a fine releguntur ad summum. He then recalls to his correspondent a stock example:
Roma tibi subito motibus ibit amor
That the scepticism attached to the Rebus was justified seemed at first confirmed by the recent Hungarian inscription from Aquincum.16 Here the Rotas square is preceded by two words Roma tibi on one line, and the letters ta on a second. Szilâgyi read sub to the right of tibi, supposed that ta was a mistake for to, and concluded that here was an example of the very versus recurrens mentioned by Sidonius (Roma tibi subito). If this were so, it would be probable that the Rotas square also is simply a versus recurrens and devoid of any cryptic meaning.
Later inspection of this inscription by Marichal and Carcopino17 has, however, shown that this first reading was superficial and that the line should actually read Roma tibi salus ita; i.e., these twenty-five letters enshrine a promise or hope of salvation which might well be appropriate for a Rome torn by imperial dissension or menaced by barbarians.18 The function of the Rotas charm immediately following might therefore be to secure this salvation. When it is further noted that the letters P and R in the square are barred in the manner of a Greek tau (a well known symbol at this time of the Cross) and that the inscription occupies a segment of a large cross of St. Andrew on the tile (this cross in the form of a chi-iota was symbolic to Christians of their saviour), it seems fairly obvious the Rotas rebus here is not merely a palindrome, but of some deeper significance.
If the sceptics have been unfortunate in their attempts to explain away the formula, little success has compensated the faithful for their credulity. Jules Quicherat19 suggested that the lines should be read boustrophedon, i.e., as the plough turns – left to right, right to left. This results in a reduplication of the same verse – SATOR OPERA TENET – TENET OPERA SATOR – which might conceivably be translated “As ye sow, so shall ye reap.” The main objection to this solution is that it entails the double use of TENET. Again, though the Sator square can be successfully read boustrophedon, there is a further difficulty with the Rotas square, whose first line must be read right to left, a procedure contrary to the rules of writing boustrophedon. A similar suggestion has been made by Grosjean,20, who reads sat orare poten, i.e., Can you pray enough? The difficulty with this solution is that it is unfeasible in the case of the Rotas square, i.e., what is historically the earlier version of the Rebus.
Most attempted solutions have foundered on the word AREPO. One method has been to treat these five letters as initials or abbreviations. Some of the monstrosities this has produced are: SA(LVA)TOR A RE (X) P(ONTIFEX) O or SATOR A R(ERUM) E (XTREMARUM) P(RINCIPIO) O(MNI).21 Alternatively, Haverfield & Collingwood22 treat the word as a proper noun, though admittedly one of no known connotation. Carcopino23 also treats AREPO as ἅπαξ, but believes that the word is Celtic in derivation and means ‘plough’. This etymology is based primarily on a remark of Columella,24 an agricultural writer of the first century A.D., who says that the Gauls call half an acre arepennis. Further evidence comes from Pliny,25 who described the invention in Gallic Raetia of plaumorati, a new kind of plough whose principal feature was two small wheels or rotulae. Carcopino argues that ROTAS in the square is connected with rotulae, while AREPO is the Gallic plough. Confirmation of this theory is sought in a fourteenth-century Greek Bible26 where a Sator square is translated into Greek: ὁσπεἱρων ἄροτρον ϰραεί ἔργα τρόχος. Here AREPO is translated by arotron, meaning plough. Any weight this might have carried is unfortunately undermined by the writer’s faulty knowledge of Latin, if not of Greek. Here opera, which must surely be, if anything, the ablative singular of opera, operae, is translated as though it were the accusative plural of opus, operis. AREPO also is apparently rendered by the accusative, making – along with τρόχους (meaning wheels) – a grand total of three accusatives in one sentence.27 In any case, if AREPO really meant plough, one might justifiably ask what on earth the sower is doing with the plough.28
The anagrammatic method of rearranging the letters of the square has provided a different approach. Attempts of this kind fall into two groups, those which have produced a fervent, pious prayer, and those which have revealed incantations to the Devil. Credible examples of the former are a formula for exorcism:
RETRO SATANA, TOTO OPERE ASPER, and the prayers:
ORO TE PATER, ORO TE PATER, SANAS
O PATER, ORES PRO AETATE NOSTRA
ORA, OPERARE, OSTENTA TE PASTOR
To the second category belong the spells of black magic:
SATAN, ORO TE, PRO ARTE A TE SPERO
SATAN, TER ORO TE, OPERA PRAESTO
SATAN, TER ORO TE, REPARATO OPES.29
Perhaps the most ingenious of these anagrams is that of a German, Kuno Von Hardenberg,30 who believed he had discovered in the square a reference to the comfort the Rose of Sharon is said to have brought to St. Peter for his sin in denying Christ. PETRO ET REO PATET ROSA SARONA; i.e., ‘For Peter even guilty the rose of Sharon is open.’ Unfortunately, the authority given for this incident, Acts 9. 35, is dubious, and there is no reference to the Rose of Sharon, at least in the Vulgate. We must conclude therefore that this incident is apocryphal (possibly a poetic tradition) and as suspect as the Latinity of Von Hardenberg’s solution. An equally impossible answer is that of Kolgerg,31 who simplified his task by having recourse to abbrevi¬ations and calmly deduced from the twenty-five letters of the square the thirty-six letters of the monastic rule. SAT ORARE POTEN (TER) ET OPERA(RE) R(ATI)O T(U)A S(IT).
Although none of these solutions are of more than moderate credibility, it was clear that some convincing Christian explanation might eventually be found. Then, in 1924, C. Frank32 made the startling discovery that the square could be so arranged as to produce the first words of the Lord’s Prayer twice over (except that there was only one N instead of two), plus two A’s and two O’s. Shortly afterwards, Grosser33 came to the independent conclusion that this unique combination could be explained by a cruciform arrangement whereby the N was used twice. The remaining four letters, two A’s and two O’s, would then be disposed thus:
A
P
A
T
E
R
APATERNOSTERO
0
S
T
E
R
0
A theory such as this cannot be proved. Its strength lies in its intrinsic probability, plus the fact that the mathematical odds against such a combination occurring by chance are astronomical. Frank was subsequently supported by Jerphanion, who in an ex
haustive investi¬gation of the square’s origin and history added a rider.34 He reported that a correspondent had pointed to the position of the T’s, which are in every case flanked by A and O.
ROT AS
O E A
TENET
A E O
SATOR
The first literary reference to the use of the T as a symbol off the Cross is in an obscure passage in the Epistle of Barnabas.35
Learn therefore children of love concerning all things abundantly, that Abraham, who first appointed circumcision, looked forward in the spirit unto Jesus, when he circumcised, having received the ordinances of three letters. For the scripture saith: “And Abraham circumcised of his household eighteen males and three hundred.” What then was the knowledge given unto him? Understand ye that he saith the ‘eighteen’ first, and then after an interval ‘three hundred’. In the eighteen, I stands for ten, H for eight. Here thou hast Jesus (ΙΗΣΟΥΣ). And because the cross in the T (= 300) was to have grace, he saith also ‘three hundred’. So he revealeth Jesus in the two letters, and in the remaining one, the Cross.
The A/0 sign does not appear in inscriptions until the end of the third century; but scriptural authority for its early use as a Christian description of God the Father and of Christ can be found in three passages of the Apocalypse.36
It might be thought that with the recognition of so many Christian symbols in combination the mystery of the Rotas square was solved. Grosser, unaware of the Cirencester square which Haverfield claimed to be Roman, had suggested that the formula originated at some time in the period before the Peace of the Church, and predicted that Roman examples would soon be found. The publication by Rostovtzeff of the examples from Dura-Europus seemed to fulfil this prophecy, and the majority of scholars, hastily recanting from their previous heresies, were now converted to the Christian interpretation of the square’s origin.
The discovery at Pompeii, however, of specimens reasonably conjectured to be earlier than A.D. 79, together with the implied presence of Christians before the eruption of Vesuvius, presented grave difficulties. These were formulated by Jerphanion at a meeting of the Académie des Inscriptions in 1937.37 (a) It is improbable, though by no means impossible, that there were Christians at Pompeii before it was destroyed. (b) The configuration of the intersecting PATERNOSTER’s presupposes that the Cross was already a Christian symbol before A.D. 79. This usage is not otherwise known before the Epistle o f Barnabas, whose date of composition is probably A.D. 130-131. (c) If the square had been invented by Christians of the first century, it ought to have been in Greek, since Greek rather than Latin seems to have been used for teaching and liturgy. (d) The Christian use of A and O was inspired by the passages in the Apocalypse, which in A.D. 79 ‘n’était pas écrite.’ (e) Cryptic Christian symbols first appear during the persecutions of the third century.
In the face of these difficulties, Carcopino38 has argued that the Pompeian examples were, in fact, written after the eruption by treasure seekers burrowing among the ruins. Della Corte, himself, describes the evidence for these early excavations,39 and in the house of Popidius Priscus40 an inscription in rough letters on the right wall of the vestibule, reads Δομμος Περτουσα, i.e., domus pertusa.41 No one would deny that this graffito was written after the eruption; but the same is not true of the graffito bearing the undamaged Rebus. This was neatly inscribed on the plaster of a column of the Palaestra. This building lies away from the better-class houses, where clandestine scavengers would be most likely to dig. In any case, the undisturbed nature of the ground precludes their presence here. It should be explained that, when Vesuvius erupted, a layer of fine ash was deposited, which covered the bodies of men and animals who were asphyxiated and buried. Above this ash is a second stratum of small stones and dust, where it would be easy to detect crude digging. Any treasure hunter wishing to write the graffito would have had to penetrate both strata, a disturbance which could hardly have escaped the notice of the excavators directed by Della Corte.
If, then, the Pompeian squares were inscribed before A.D. 79, some alternative solution must be found. Could they, in fact, have been the work of Christians? There were certainly Christians in Rome. Tacitus, speaking of the disturbances of A.D. 64, when the fire of Rome touched off the first great persecution, refers to Christians as an ingens multitudo.42 This may be rhetorical exaggeration, but there is clear evidence of quarrels between Christians and Jews under Claudius.43 Again, in his Epistle to the Romans, St. Paul speaks with awe of the Church at Rome as a comparatively old institution.44 We know, too, that St. Paul once stayed at Puteoli, only a few miles from Pompeii.45 There is ample evidence of communication between the two towns (e.g., a graffito found at Pompeii recording greetings to the colony of Puteoli46 and it is hard to believe that news of St. Paul, or of the Christians, had never travelled from Puteoli.
Some archaeological evidence may possibly support these general considerations. In the town of Herculaneum, which is adjacent to Pompeii and was also destroyed in the eruptions, excavations have unearthed a large, two-storey house of about A.D. 50. A panel of stucco in an upper apartment has been discovered bearing the imprint of a cross.47 Traces of nails suggest that the cross was of wood, while other nails in the panel may mean that the cross was removed and a cover placed over the area. With this may be compared a similar cross on a panel of white stucco from Pompeii. This was published by Mazois in 1824, along with a celebrated inscription written with charcoal in the atrium of house no. 22.48
This evidence, however, is tenuous in the extreme. Crosses do not necessarily imply the presence of Christians,49 and the character of the find spot at Herculaneum, with its wooden dice box and loose die, has been held to rule out any religious connection. The cross at Pompeii has never been accepted as genuine, and when the original charcoal inscription faded, sceptics were quick to emphasize discrepancies in the two copies of the original.50 In the present state of the archaelogical record, the most that can be said is that there may have been a few, solitary Christians in the area. There is certainly no justification for supposing the existence of a Christian community, and it need hardly be stressed that their presence at Pompeii would offer no solution to the remaining difficulties of Jerphanion.
A pagan origin for the square is also unlikely. It has been argued by Sundwall51 that it has its roots in the Orphism popular in South Italy. On this interpretation, the Sower is Triptolemus, while the wheels symbolize the plough, which is one of his attributes. An Italian origin is also suggested by Eitrem,52 who points to the central N marking the nave of the wheel whose four spokes are NET.
T
E
TENET
E
T
The meaning would then be NET (neoI spin) OPERA ROTAS (= rotans), i.e., “She (a deity, demon, or the inscription itself) spins her works revolving.” Neither of these theories gives a convincing explanation of the complete square, and it should be noted that both abandon Grosser’s important discovery. As has already been stated, the odds against the fortuitous occurrence of the double PATERNOSTER with the A’s and O’s are extremely high.
We are left with the probability that the Pompeian examples are Jewish. Large numbers of Jews had, in fact, been settled in Pompeii53 and its neighbourhood in 62 B.C. after Pompey’s campaigns in the East. Their reputation as superstitious charlatans and dabblers in magic had been widespread since the days of Moses,54 and they were notorious for their use of magic talismen, amulets, spells, and riddles.55 Word magic, alphabetic acrostics, and gematria, by which a numerical value was ascribed to the individual letters of a word, played an important part in Jewish exorcism, cosmogonic theories, and the symbolic representation of divine powers.56 Not only were the letters of the alphabet believed to comprehend all knowledge, but the written word, in particular, was held to be charged with magic.57 Hence the efficacy of the palindrome, whose magic could not be destroyed whichever way the spell be read.
A rebus which is typical of this magic genre may well have been inscribed by Latin-speaking Jews, familiar with Hebrew and the Hebraic method of writing. Such a solution would also provide a convincing answer to the difficulties inherent in a Christian origin. Although the A/0 sign may not have been in Christian use before its appearance in the Apocalypse,58 the idea occurs much earlier in such passages as Ex. 3.14; Is. 41.4, and 44.6. The letters aleph and thau are also used in the Talmud to symbolize completeness and totality.59 The appearance of this symbolism in a Jewish rebus would therefore be quite plausible, particularly in conjunction with the T’s of TENET.60 These may best be explained not as Christian crosses, but as a Latin form of the Jewish thau sign, the symbol of salvation which, in the vision of Ezekiel, saved the Just from the avenging angel.61 This mark in its archaic form (+) appears regularly on ossuaries of the Hellenistic and early Roman periods, particularly at Jerusalem.62 The central position of the N may also be significant. The Jews attached a peculiar power to the mere pronouncement of the ‘Name’, in particular of the Divine Name.63 As the initial letter of the word nomen, N might also serve here as the Latin equivalent of the Hebraic םש, the unique Divine Name, fount of Divine Power and centre and origin of all things.64
A Jewish origin also provides a satisfactory explanation for the resolved form of the Rebus, the Paternoster amulet. Not only are the positions of the A’s and O’s and the central N even more striking, but the overall configuration is clearly an archaic thau. Most signifi¬cant of all is the use of the Pater Noster. Far from being a Christian innovation, this form of address has its roots in Judaism.65 In the Babylonian and Palestinian recensions of the Schemone Esre66 God is frequently addressed as ‘Our Father’,67 and a similar invocation to ‘Our Father’ is found in the prayers of Rabbi Eliezer (died ca A.D. 90) and Rabbi Akiba (died ca A.D. 135).68 It seems possible, then, that the Pater Noster invocation was as familiar to the Latin-speaking Jews of Pompeii as to any (hypothetical) Christians.
The remaining question is the proper interpretation of ROTAS OPERA TENET AREPO SATOR. Despite efforts to trace its imagery to the Ezekiel passage69 or to the concept of logos as charioteer,70 no explanation has ever convincingly elucidated its Delphic meaning. It is doubtful whether such attempts are justified or legitimate. These are simply five words ingeniously evolved from the Paternoster charm in such a way that, when properly combined, they form a square which can be read in four, different directions. The ‘magic’ of the square is basically the perfect symmetry of its component letters. These also betray cryptic Jewish symbols to those who know their origin and secret. To construct such a square from the Paternoster amulet is a technical achievement of the highest order. To require that the individual words, one of which, the palindrome of OPERA,71, is not even a Latin word, will also be meaningful when read concurrently72 is to expect the impossible. Any superficial meaning therefore which may be allegedly wrung from them is purely accidental.
The final verdict on the origin of the Rotas-Sator square is clearly dependent on future archaeological discoveries. But in the present state of the evidence it seems reasonable to conclude that this charm, at least in the form we now have it,73 originated with Latin-¬speaking Jews (presumably settled in Italy) in the period immediately prior to the Christian Era. Such an origin is itself sufficient explanation of its cryptic form; alternatively, it may have been a product of the pogroms of, e.g., A.D. 19 or A.D. 49. It would seem that it fell into disuse, to be revived later as a Christian symbol amid the new enthusiasm for symbolism characteristic of the third century and later.74 Perhaps its obscurity in the meantime explains why its origin was forgotten and its Christianity so readily accepted.
