Part I

THE question may now be asked, how was it made possible to read the wedge-shaped characters found on the monuments, votive offerings and the tablets? When Pietro della Valle ^[1] brought specimens of the writing to Europe, it was the first time that such characters which did not resemble any known alphabet were seen by European scholars. They seemed so strange that it is scarcely surprising that some scholars ^[2] questioned whether they stood for real writing or were not merely ornamental decorations. The excavations in the mounds of Babylonia and Assyria proved conclusively that cuneiform or wedge-shaped writing was the general and only script used in the region.

The strange script, however, was not limited to Babylonia and Assyria. In fact, the first specimens to reach Europe, copied, as we have seen, ^[3] by Pietro della Valle, came from a place that lay outside of the Euphrates Valley, and it was on cuneiform inscriptions of this type that the first attempts at decipherment were made. Delia Valle, as well as other travellers, had passed in their travels the chief sites of the old Persian Empire and were particularly impressed by the tombs and the remains of great structures still standing at Persepolis, the name of which ("Persian city") preserved the tradition that it was one of the political centres in the days of the great Persian kings. Here at least there was a definite starting-point.

If cuneiform inscriptions were found on monuments erected by Persian rulers, the conclusion was obvious that the characters represented the ancient Persian language which was the official speech of the empire. At Persepolis it was not necessary to dig below the surface to come across remains of Persian days. The ruins of a great palace were still standing. ^[4] A large number of high and beautiful columns were still in place, and by their help one could trace the general divisions of the structure of which they formed a part. Besides the columns and portions of walls, sculptured monuments of various kinds were scattered about, besides magnificently decorated tombs cut in the rocks that surrounded the city.

These well preserved monuments were covered with the wedge-shaped characters. An English traveller, Herbert, ^[5] was among the first, towards the close of the seventeenth century, to give an account of the strange writing which he (like della Valle) correctly conjectured was to be read from left to right, and he also concluded that the writing represented the language spoken by the Persians. He despaired, however, of the hope of the writing ever being deciphered unless (as he says) by another Daniel who was able to read the mystic writing on the wall of Belshazzar's palace.

In 1711 the first complete inscription from Persepolis was reproduced by a French traveller, the Chevalier Chardin, ^[6] from which it should have been evident that although the characters always had the form of wedges, still the combinations (varied considerably and that there were in reality three quite distinct styles of cuneiform writing on the rocks and monuments of Persepolis. Although a number of intrepid travellers and careful observers like Engelbert Kaempfer ^[7] and Cornells de Bruin ^[8] examined the inscriptions, it was not, however, until the second half of the eighteenth century that Carsten Niebuhr, whom we have already come across ^[9] and who copied more of the inscriptions than any of his predecessors, recognized the fact of three distinct varieties of the cuneiform characters at Persepolis, varying in the complexity of the combinations of the wedges.

PLATE XIX

Fig. 1, Ruins at Persepolis

Fig. 2, The Propylea of the palace of Xerxes I (476-465 B.C.) at Persepolis

Though distinguishing these three varieties as Classes I, II and III, Niebuhr did not draw the further conclusion that the varieties represented three distinct languages, but supposed all three to be the same language, written in a threefold form. He even correctly analyzed the characters in Class I as consisting of forty-two signs and concluded that this form represented an alphabetic method of writing. ^[10] On the basis of Niebuhr's work, two scholars who were trained philologists proceeded to make the first attempts at decipherment.

Tychsen ^[11] drew the correct conclusion that the three varieties represented three distinct languages. He furnished a tentative translation of one of the smaller inscriptions of Class I which, however, was pure guesswork, and turned out to be entirely erroneous, except for the fact that he correctly assumed a certain character to represent the vowel a.

Tychsen proceeded on the erroneous assumption that the buildings and inscriptions at Persepolis dated from the late Persian dynasty, known as the Parthian, in the third century of our era. But for this error, he might have made further progress in the decipherment. The correct identification of the remains at Persepolis with the Achaemenian kings of Persia in the sixth and fifth centuries before our era was made by a contemporary of Tychsen, Prof.

Friedrich Miinter of Copenhagen, who instituted a comparison between the monuments at Persepolis and those at Naksh-i-Rustam, which the researches of a famous orientalist, Sylvestre de Sacy, had shown to be the tombs of kings of the Arsacidian dynasty. The result was to establish the identity o,f the art at Persepolis and Naksh-i-Rustam, further reinforced by the occurrence of the same fabulous animals or symbols on the monuments of both places.

The art was distinctly Persian, as were the costumes and ornaments on the figures at Persepolis. Miinter made some further progress also in unraveling the mystery of the inscriptions. He recognized that a diagonal wedge occurring constantly in the inscriptions of Class I was a word separator, a clue that proved to be of the greatest possible value, since it enabled scholars to definitely fix the beginning and end of each word. Another suggestion thrown out by Miinter, that a series of seven characters occurring in all inscriptions stood for the word king, was finally rejected by him, though the conjecture proved to be correct.

Fortunately, not long before the time that Tychsen and Miinter were groping their way in the dark, a French scholar, Anquetil-Duperron, was busy in the East collecting manuscripts of the Avesta, the sacred writings of Zoroastrianism, and through native Parsi priests was learning how to read the characters and to interpret the contents of the sacred books. ^[12]

Through the publication of his material, scholars had before them specimens of the language employed in the days of the Persian rulers. The characters used in the Avestan manuscripts were, however, totally different from those found on the Persepolitan inscriptions ; they represented a cursive alphabet that probably had its origin in India and was adapted to the old Persian language. To be sure it subsequently turned out that the Avestan books represented a compilation covering a long period of gradual growth and that even the oldest portion could not be earlier than the fourth century, while the introduction of the Avestan alphabet could not have taken place before the third century.

We were, therefore, still some distance from the time of the earliest Achsmenian rulers, but close enough to warrant the assumption that the language of the Avesta was practically identical with that spoken by Cyrus and his successors. The task of scholars, therefore, lay in attempts to recognize in the wedge-shaped characters the consonants and vowels corresponding to the signs for these in the Avestan alphabet. There was, of course, no possible connection between the forms of the Avestan and the cuneiform alphabet, but the same sounds must be represented in both, and the words spelled out in the Persepolitan inscriptions must be close enough to such as were furnished by the Avestan writings to show that they were genuine Persian words.

The problem, therefore, resolved itself into a species of rebus, somewhat as though one were to write English with Sanskrit characters and then to determine by patient endeavor the value of the Sanskrit characters so as to furnish good English words, and above all, a sequence of thought. Simple as this may sound, it involved great difficulties because of the imperfect knowledge, at the end of the eighteenth century, of the Avestan language, the study of which was still in its infancy, and because of the puzzling circumstance that Class I of the Persepolitan inscriptions showed fortytwo characters too many if each combination of wedges represented a single sound, too few if the method of writing was syllabic, ^[13] and not alphabetical.

Now in many of the inscriptions from Persepolis it was observed that certain words occurred frequently in all of them. It could furthermore be concluded on the assumption that Class I represented the Persian of the days of the Achsemenian kings that the names of the rulers should be found on them, and with the names also the titles. The next step seemed simple enough to try to fit the sounds composing the names of the Persian kings which were known to us from the Old Testament, from Herodotus and from other sources to the series of characters in the Persepolitan inscriptions that might represent proper names.

Had Munter not rejected his conjecture that a certain series of characters stood for the word "king", ^[14]he might have been the one to take the next step and to become the decipherer of the inscriptions. Munter was led to seek for the word for king in the Persepolitan inscriptions by the analogy which they presented to those on the royal tombs at Naksh-i-Rustam. Greek inscriptions at this place by the side of those in the Pehlevi script furnished de Sacy, whom we have already mentioned, ^[15] with the clue both to the historical character of the monuments and to the decipherment of the Pehlevi script, which turned out to be a variety introduced into Persia during the rule of the Sassanian kings (227-641 A.D.).

The Greek inscriptions based on Pehlevi models, of which they were in fact translations, revealed a stereotyped order of phrases and titles on these monuments. The beginning was made with the name of a ruler followed by his titles, and these in turn by the name of his father with his titles. This gave a form as follows :

N, great king, king of kings, king of Iran and Aniran, son of N, great king, king of kings, king of Iran and Aniran.

With the help of several bilinguals Greek and Pehlevi de Sacy, through fitting the proper names on to the characters, the position of which could be determined by the place occupied by proper names in the Greek translation, succeeded in determining the characters of the Pehlevi alphabet, while as soon as he was able to read words, the practical identity with the older Persian, now revealed through the researches of Anquetil-'Duperron, furnished an unfailing aid in recognizing the meaning of the words written in the Pehlevi script. Here then a rebus was correctly solved the characters fitted on to the sounds which, since the words thus put together were Persian and gave a connected sense, were shown to be the correct ones.

Miinter availed himself of de Sacy's results as a support for his thesis that the Persepolitan inscriptions were those of the early Persian kings, but he stopped short at this point. Had he clung to his guess regarding the combination of signs representing the word for king, it would no doubt have occurred to him to apply the stereotyped form of the Pehlevi inscriptions also to the Persepolitan monuments.

This step was taken by the man who was destined to achieve immortal fame as the decipherer of the Persian cuneiform inscriptions Georg Friedrich Grotefend (born 1775), a teacher of Greek in tbe gymnasium at Gottingen, who, on the fourth of September, 1802, read a paper before the Gottingen Academy in which he claimed to have found the key to the reading of the inscriptions of Class I. The paper, ^[16] consisting of three parts, began with a general consideration of the three varieties of script on the monuments of Persepolis.

Grotefend showed the definite basis for assuming that the three varieties represented three different languages, that the variety which occupied the first place when the three scripts were written one under the other, or which was above the head of a figure the most prominent place, while the two others were grouped to either side, represented the old Persian language spoken in the days of the Achaemenian kings (539-331 B.C.). If, therefore, the first class could be deciphered, it would be possible to use the inscriptions of this class as a basis for deciphering the other two classes which must represent translations of the old Persian into two languages that were spoken by the subjects of the Persian Empire.

A parallel to such a procedure exists to-day in the decrees of Austro-Hungary which are issued in German and Hungarian. ^[17] Class I would serve as the key to Classes II and III, just as de Sacy used the Greek inscriptions at Naksh-i-Rustam to decipher the accompanying Pehlevi inscriptions the Greek being a translation of the Pehlevi ; and as in the decipherment of the Egyptian inscriptions, the Greek translation of the hieroglyphic inscription on the stone found at Rosetta served as a key to Francois Champollion. ^[18]

Specimen of the three classes of Cuneiform characters.
B and G -- old Persian (class I);
C -- Babylonian-Assyrian (class III);
D -- Neo-Elamitic (class II)

Grotefend also confirmed the results reached by his predecessors that the order of the writing in all three varieties was from left to right and that the lines followed horizontally and not vertically, ^[19] or in boustrophedon fashion, ^[20] as some scholars had maintained. In one deduction Grotefend erred, though fortunately it did not affect his key which he applied merely to Class I. He maintained that all three varieties of cuneiform writing represented an alphabetical script, not therefore ideographic like the Chinese, nor syllabic like the Japanese. He was right so far as Class I was concerned, but wrong as to the other two classes which turned out to be partly ideographic and partly syllabic.

Coming in the second part of his paper to the inscriptions of Class I, he picked out of the forty-two characters comprising the script, eight which occurred with great frequency (and two or more of which appeared in every word) and concluded that they were vowels. Availing himself at this point of the stereotyped form of the Pehlevi inscriptions of Naksh-i-Rustam and concluding that the Sassanian rulers followed in this respect the model of the old Persian kings whose realm they had taken over, he proceeded to pick out the proper names in the Persepolitan inscriptions of which one ought therefore to be found at the beginning and another somewhat further on the name with which the inscription began being that of the one who is commemorated by the monument, and the other the name of the father.

In the third part of his paper he took up two of the short inscriptions that had been copied and published by Niebuhr and which, in his publication, were numbered B and G, the former consisting of six lines, the latter of four lines. The analogy with the Pehlevi inscriptions led him to look for the word king, which ought to follow the name at the beginning of the inscription, and should appear several times even in a short inscription. The diagonal wedge which Miinter had conjectured to be a word separator, made it easy to pick out a series of characters constituting a word, and it was not long before Grotefend hit upon seven characters occurring just where one Mould expect the word for king.

These signs were:

From the dictionary which Anquetil-ODuperron had compiled for his Avestan texts, the word for king was given as Khsheio. ^[21] Now some of the seven characters composing the word for king occurred in the series of characters that constituted the first word in each of the two inscriptions, and which, on the analogy of the Pehlevi, ought to be a proper name. Grotef end's next task, therefore, was to study the characters composing these two names carefully. They must, of course, conceal the names of old Persian rulers, known tcxus, as has been indicated, ^[22] from various sources. Grotefend observed that in the inscriptions of Persepolis at his disposal, there were only two varying series of characters constituting the beginning of the inscriptions, which meant that all the inscriptions proceeded from two rulers or, at all events, from rulers whose names alternated between X and Y.

Taking up now the two inscriptions B and G selected by him, he found the proper name in B to consist of seven characters as follows:

and in G also of seven characters

He further noted in G the occurrence of the same name with which B begins, forming, to wit, the second series of characters in the third line. ^[23] The analogy with the Pehlevi inscriptions made it certain that the king represented by this series of characters was the father of the one whose name appeared at the beginning of G. In other words, if we designate the proper name with which inscription B begins as X, and the one with which G begins as Y, we would have the relationship

Y is the son of X.

Both X in inscription B, as well as Y and X in inscription G, were followed by the same series of characters that had been conjectured to represent the word for king. Grotef end could therefore go a step further and read in G

Y king .......... son of X king.

The three characters following the group X king in the fourth line of G (after the word separator) Grotefend, after the analogy of the Pehlevi inscriptions, assumed to be the word for son. Searching for these signs in inscription B, he found them in line five immediately after the word separator, which closed the word running over from line four. ^[24] Hence he concluded that the name of X's father should be found before these signs for son. It was at this juncture that his keen ingenuity showed itself. In G the name of X, the father of Y, was followed by the characters for king. He could not find these characters in line four of inscription B and therefore concluded that the ten characters preceding the series "son" in B (beginning in line four and running over into line five) represented the name of X's father, and that this personage was not a king. This could only mean that X was the founder of a dynasty not, therefore, himself of royal descent. If we call this group of ten characters Z, we would therefore have two series :

In inscription B: X king ..... son of Z.
In inscription G: Y king ..... son of X king,

which gives us the order

Z, X, Y = grandfather, father, son.

Now in the lists of the rulers of Persia ^[25] there were during the first period two dynasties, (1) the one founded by Cyrus who was succeeded by his son Cambyses (539-522 B.C.), (2) a second founded by Darius I, succeeded by his son Xerxes I and grandson Artaxerxes I (522-424 B.C.). Then came an usurper, Xerxes II, who ruled for forty-five days, followed by Darius II, whose son, grandson and great-grandson followed one another, Artaxerxes II, Artaxerxes III and Arses (424-336 B.C.), after which came Darius III (336-331 B.C.), who succumbed in 331 B.C. to Alexander the Great. Grotefend thus had a choice for Z, X, Y between

(a) the father of Cyrus, Cyrus himself and Cambyses, or (&) Hystaspes (the father of Darius), Darius himself and Xerxes, or (c) the father of Darius II, Darius II and Artaxerxes II. The first case was ruled out by the circumstance that the names of Cyrus and Cambyses began with the same letter, whereas X and Y began with different characters. X and Y would therefore turn out to be either Darius and Xerxes, or Darius II and Artaxerxes II. The latter alternative, if correct, would involve that the name Y as the son of X should be longer than X. Both X and Y, however, consisted of the same number of characters, namely seven.

Grotefend was thus thrown back on the second of the three possibilities, Darius and Xerxes, as the one fitting the required conditions. Taking up his three names once more, he observed that the first two characters of Y corresponded to the first two characters in the series that represented the word "king", which being, as we have seen, Khsheio, showed that these two characters must stand for the sound kh and sh respectively. King Y therefore bore a name beginning with kh.sh. The fourth and the seventh character in this name was the one which, because of its frequency, Grotefend assumed to represent the vowel a or e, while the sixth sign was again the sign sh.

He thus could read the name partially as

Completing the word for king in the series of seven characters, he assigned values tentatively as follows:

(3rd sign) ^[26], (4th sign) ^[27], (5th sign) ^[28]

which was certainly close enough to the form Khsheio to justify Grotefend's confidence in his method. Proceeding to compare the signs composing the name of King X with those in the word for king, he could read three of the seven characters as follows:

Now the Old Testament form of the name Darius (occurring in the Book of Daniel), which would be nearer to the original pronunciation of it than the Grsecized form Dareios, was Daryawesh. It was, therefore, like fitting on a rebus to assign the value d to the first sign and r to the third, which gave him

d.a.r.h. . . .sh

All that was needed was to assume the value e and u for the fifth and sixth characters to obtain the full name

The assigning of the value r to the third character was confirmed by the occurrence of the same character as the fifth in the name Y, while the value h occurring as the fourth character in the series for "king" and in the name X also fitted in with Y, though it ultimately proved to be erroneous. He thus could read Y as

(4th sign), ^[29]

which could very well be the original form of a Persian name distorted by the Greeks to Xerxes. ^[30]

Passing to the name of the father of Darius, his decipherment of Darius and Xerxes gave him the values of the third, fifth, ninth and tenth characters as

The eighth because of its frequent occurrence he assumed to be also the vowel a. ^[31] The Greek form of Darius' father was Hystaspis, but several other forma were known from various sources, including Goshtasp or Gushtasp and Vishtaspo. Grotefend assumed the original consonant to have been g ^[32] and, accordingly, he supplied the values for the remaining characters as g for the first, o for the second, t for the fourth, s for the sixth, p for the seventh, and thus obtained the reading

He had thus succeeded in puzzling out three proper names and the word for king; he could feel tolerably certain that he had correctly identified the two kings, Darius and his son, Xerxes, as the authors of the two inscriptions. It was subsequently shown that he had erred in a number of the values assigned by him to the fourteen signs, but the way had been opened for further progress.

Taking up the two inscriptions B and G for further comparison, he noticed that the third word in both was the same. The analogy with the stereotyped form of the Pehlevi inscriptions suggested that this word was an adjective like "great", descriptive of the preceding word king. The fourth and fifth words likewise agreed in both B and G.

The former was again the word for king, while the latter was king, plus four signs, which indicated some form of this word. The stereotyped form of the Pehlevi inscriptions read: X king great, king of kings. Grotefend assumed the same model for the Persepolitan inscriptions, which enabled him to read B as

Darius, great king, king of kings

and G as

Xerxes, great king, king of kings.

Then followed in G the name Darius with the word king, and thirdly, the series of signs which, occurring in both inscriptions, he had assumed to be the word for son, with the preceding name in the genitive. ^[33] He thus could read all of G except the last word, as

Xerxes, great king, king of kings,
son of Darius king, ^[34]

In B there were five words which he could not determine by this process.

Darius, great king, king of kings,
king ^[35] ..... son of Goshtaspa (i.e., Hystaspis)
..........^[36]

As for the values of the word for son and the plural of the word for kings, some of the signs had already occurred in the proper names. So in the three signs

which he had identified as son, the middle one was the character to which, in the name of Darius, he had given the value u. Prompted by the existence of a word "bun" in Pehlevi in the sense of offspring, his guess of 6 or p for the first sign of the word for son was correct, ^[37] his conjecture for the third as n was far off, since the correct reading was shown by Sir Henry Rawlinson, in 1847, to be tr. Similarly his supposition for the second and fourth signs of the ending added to king indicating the plural as s and o were both wrong. A Danish scholar, Rask, in 1826, correctly determined the values as n and m respectively. Nor did Grotefend have more success in his attempt to read the additional words in B or the last word in G. There was thus still much left to do before it could be said that a firm basis had been secured for the decipherment of the Persepolitan inscriptions.

Grotefend's method had been successful in determining with tolerable certainty the reading of the three proper names, but when he attempted to read other words, he floundered about and generally went wrong. The errors made by Grotefend were seized upon as a basis of attack, and only scant acknowledgment was made of his success in identifying the three proper names.

The Gottingen Academy published merely an extract of his attempt at decipherment in 1802, ^[38] and it was not until three years later that a fuller account of his decipherment appeared as an appendix to Heeren's work on the "Politics, Intercourse and Trade of the Principal Nations of Antiquity", ^[39] Meantime Grotefend had profited by some of the criticism passed on his efforts and had succeeded in reading the last word in inscription G, identical with the second in line five of B as Achsemenian, so that he could read this inscription completely as

Xerxes, great king, king of kings, son of Darius, king, the Achæmenian.

To have thus determined the ancient designation of Achsemenian by which all the Persian rulers from Cyrus down were known was a considerable step in advance.

A striking confirmation of Grotefend's identification of the name of Xerxes was furnished by Saint-Martin, in 1823, who took up the fourfold inscription on an alabaster vase, published as far back as 1762, ^[40] and which, as now was apparent, contained an inscription in the three classes of cuneiform script, besides an Egyptian inscription enclosed in a cartouche;. The Egyptian name had been read by Champollion as "Xerxes, great king", and Saint-Martin showed that the cuneiform inscription in Class I tallied completely with the signs read in inscription G by Grotefend as Xerxes, followed by the words for "king" and "great".

In this way there was established a mutual confirmation of the key to the reading of both the hieroglyphic and the old Persian inscriptions. Other Oriental philologists now took up the task. A Danish scholar, Rask, in a study on the Zend language, ^[41] showed its affiliation to Sanskrit, though a separate language and quite as old as Sanskrit, and that it was closely related to the language of the Persepolitan inscriptions, as Grotefend had indeed assumed.

He was able to correct Grotefend's reading of the genitive plural attached to the word for "king", and thus at one stroke definitely determined the correct value of two signs (m and n) that had not occurred in proper names. His philological training also enabled him to prove that each sign in Class I could have only a single value and not, as Grotefend supposed, more than one.

The establishment of this principle marked a forward step in determining the signs in this class that stood for vowels. Further progress in the study of old Persian, or Zend as it was at that time called, was made by the most eminent Persian scholar of his day, Eugene Bournouf , of Paris.

As a result ^[42] he could correct Grotef end's reading of the word for "great" following that of king and he determined the values of two more signs, k and z, while Saint-Martin, whose name has already been mentioned, correctly assumed that Vishtaspa was an older form than Goshtasp for Hystaspis, and correctly read the initial sign of the cuneiform form of the name as v instead of g. A further advance was signalled by the appearance of a comprehensive work, in 1836, on the old Persian inscriptions (as the monuments of Class I were henceforth to be called), by a German Orientalist, Christian Lassen. ^[43]

Examining anew the basis of the decipherment so far as it had proceeded, he confirmed the identification of the proper names as made by Grotef end, but showed that the word for "king" and for the plural, while correctly guessed by Grotefend, had been incorrectly read. ^[44] In all, seven signs were correctly determined by Lassen's researches, so that next to Grotefend, of whose identifications eleven were definitely accepted, he has a larger share to his credit than any other of the early decipherers.

A result, however, of Bournouf's and Lassen's accurate studies was to show that, while it was true, as Rask had maintained, that each sign had only one value, the reverse of the proposition that a single letter was represented by one sign was not true. Thus, three signs having the value d had been found, two for g, two for k, three for m and two for n, r and t. If the cuneiform script of Class I was an alphabetic form of writing, what could this mean 1 ? The solution of the problem was due to the combined efforts of three scholars, Edward Hincks, Sir Henry Rawlinson and Jules Oppert, ^[45] who definitely established the fact that the use of signs having the same consonantal value differed according to the vowel that followed.

So the one form of d, g, k, m, n occurred only in case the following sign was the vowel a, whereas the other sign for d and m was used before the vowel i, the third sign for d and m and the alternate one for g, k, n, r before the vowel u, while in the case of the two signs for t, one was used before a or i and the other before u.

Before leaving the subject a few words must be said of Sir Henry C. Rawlinson, who in many respects was the most remarkable of the early decipherers, not even excluding Grotefend. While in the service of the English army ^[46] in Persia, his attention was directed to the cuneiform inscriptions scattered throughout the country. ^[47] He copied some of them and began to study the strange looking characters. He prepared a list of the signs used and without even knowing of the work of Grotefend, de Sacy, Saint-Martin, Rask, Bournouf and Lassen, started in the year 1835 to decipher the proper names in the shorter inscriptions. Unconsciously he followed exactly the same method as Grotefend, and by a strange coincidence the first three names read by him were Darius, Xerxes and Hystaspis.

Fig. 1 (left), Rock Sculpture and inscription of Darius I (522-486 B.C.) at Behistun
Fig. 2 (right), Vase of Xerxes I (486-465 B.C.)

The key to the decipherment of Class I was thus actually found twice, though the credit as the pioneer belongs to Grotefend. Rawlinson realized that in order to determine the full syllabary, that is to identify all of the forty-two characters used in Class I, a study of a large number of proper names was required. He, therefore, devoted his special attention to a long inscription cut into a rock at Behistun on the high road leading from Persepolis westwards into the Euphrates Valley.

While the inscriptions found elsewhere were brief, consisting of from 4 to 7 or 8 lines, this one contained some 400 lines in each of the three varieties of the cuneiform script. The task of copying this remarkable document, on a rock that rose several hundred feet above the road, was in itself a testimony to Rawlinson's skill and endurance. The inscription proved to have been placed there by King Darius, who recorded on it in an impressive manner his suppression of rebellions, his conquests of numerous peoples and other achievements of his reign. ^[48]

There were indications that the king had erected some kind of an ascent to the rock so that passers-by might mount to see it, but all traces of such an approach had disappeared. Rawlinson had to construct a scaffold to reach the inscription and at certain portions of the rock was suspended by a rope so as to obtain as complete a copy as possible.

Here he had an inscription with several hundred names of places that could be picked out. By dint of great perseverance he managed to read and identify with the help of classical writers and mediaeval geographers a number of such names, which furnished him with the values of no less than 18 additional signs. By the year 1839 he was able to read 200 lines of the Persian text of the Behistun inscription. Rawlinson thus worked out for himself the entire syllabary of the old Persian cuneiform script through the identification of the proper names in -the Behistun inscription. ^[49]

That his conclusions, independently reached and without his ever knowing of the work done by others, agreed in almost all particulars with the results obtained through the combined efforts of Grotefend, Saint-Martin, Rask, Bournouf, Lassen, Hincks and Oppert was a guarantee of their reasonableness and helped to inspire confidence in the method pursued. After the first steps had been taken by Rawlinson, and he learned of what the scholars of Europe were doing, he entered into correspondence with them, more particularly with Bournouf and Lassen.

Rawlinson availed himself of the results of Bournouf's researches which revealed the character of the Zend or old Persian language in such a manner as to be capable of being utilized in the decipherment of the Persian inscriptions. The comparison with Sanskrit was another aid that secured valuable results, for after the real relationship of Zend to Sanskrit had been determined ^[50] the Sanskrit could be used to settle the meaning of words where Zend failed.

By such methods the guesses and conjectures of the earlier decipherments were subjected to reliable tests and were confirmed, rejected or modified as the case might be. The appearance of Rawlinson's papers in the Journal of the Royal Asiatic Society, ^[51] therefore, placed the decipherment on an absolutely sure foundation. The "rebus" stage had been definitely passed, and it merely remained for the successors of Rawlinson to modify some of his results in minor points.

In 1862 Prof. Friedrich Spiegel's work on the old Persian inscriptions, ^[52] giving the text and translations of all the inscriptions of Class I and accompanied by a grammar of the language and a glossary, set the coping stone to the structure that had been so laboriously constructed through the combined efforts of some of the best scholars of the day.

Footnotes and references:

[1]:

Above, p. 12, seq.

[2]:

Among them the famous Thomas Hyde, Eistoria Beligionis veterum Persarum, etc., (Oxford, 1700), p. 527, S. S. Wi/tte, Professor at Rostock University as late as 1799.

[3]:

Above, p. 12. See Plate XIX, Fig. 1.
 

[4]:

See the illustrations in Stolze, Persepolis (Berlin, 1882).

[5]:

In his Some Years' Travels into Divers Parts of Africa and Asia the Great (London, 1677), p. 141, seq.

[6]:

Voyages de Monsieur le Chevalier Chardin, en Perse et autres Lieux de l'Orient (Amsterdam, 1711), 3 vols.

[7]:

He embodied the results of his travels in a Latin work published in 1712, with a long title, Amaenitatum exoticarum politico physico-medicarum fasciculi quinti, (Lemgo). Kempfer was the first to apply the term cuneiform ("wedge-shaped") to the characters.

[8]:

Voyages de Corneille le Brun par la Moscovie, en Perse et aux Indes Orientates (Amsterdam, 1718), 2 vols. [French translation from the Dutch edition of 1714.]

[9]:

See above, p. 13.

[10]:

In Vol. II of Carsten Niebuhr's Reisebeschreibung nock Arabien und andern umliegenden Ldndern, completed after his death by his son (1774-1837, Copenhagen, 3 .vols.), will be found his account of his investigations of the monuments of Persepolis.

[11]:

Olav Gerhard Tychsen, De cuneatis inscriptionibus Perse-politanis lucubratio (Rostock, 1798).

[12]:

See the account of the beginnings of the history of Avestan studies in Darmesteter's Introduction to his monumental work, Le Zend-Avesta (Paris, 1892-1893), 3 vols.

[13]:

By syllabic is meant the use of a sign to indicate an entire syllable; thus ra-shun-al would be syllabic writing, whereas r-a-t-i-o-n-a-l is alphabetic, while if some picture or a sign derived from a picture were used to convey the idea of rational, the writing would be ideographic. The sign for dollar is ideographic writing.

[14]:

See above, p. 66.

[15]:

Memoires sur diverges Antiquites de la Perse (Paris, 1793).

[16]:

The title was Prcevia de cuneatis quas vacant inscriptionibus Persepolitanis legendis et explicandis relatio.

[17]:

Some of the decrees of the Turkish Empire are similarly issued in two languages, Turkish and Arabic.

[18]:

See Steindorff in Hilprecht's Explorations in Bible Lands, p. 629 seq. The Rosetta stone also contained a version in the late demotic script of Egypt.

[19]:

Like the Chinese.

[20]:

One line from left to right, the next from right to left, and so alternately, as in the case of Hittite inscriptions.

[21]:

Be it noted once for all that kh and sh in the transliteration of oriental languages represent a single sound and in a scientific transliteration are given as h and s. For the sake of convenience I retain kh and sh.

[22]:

Above, p. 68.

[23]:

There was, to be sure, one character the sixth not found in the series at the beginning of B, but Grotefend at once concluded that this variation was due to the fact that the name in G stood in the genitive, whereas in B in the nominative.

[24]:

The Persian scribes, depending upon the diagonal wedge as the word separator, did not hesitate to allow a word to run over from one line to the other.

[25]:

See the list in Jastrow, Aspects of Belief and Practice in Babylonia and Assyria, p. 448.

[26]:

Grotefend, as occasion required, assigned *he value a or e to this sign. It turned out that the value was always a.

[27]:

This was subsequently shown to be ya.

[28]:

Subsequently shown to be t.

[29]:

Or a. Note that in this name the third sign has an additional horizontal stroke and that the fifth sign lacks a third horizontal stroke due to slight defects in Niehuhr's copies which were easily recognized as such by scholars.

[30]:

In Babylonian inscriptions the name appears as Khishiarshi, which is quite close to the correct Persian form Khshayarsha.

[31]:

It turned out to be h.

[32]:

It turned out to be v, the form Vishtaspo, found in Zend literature, being closer to the original than the late form Gushtasp or Goshtasp. The three additional signs (8, 9, 10) a, h, e, Grotefend regarded as an attached ending.

[33]:

Indicated by the three additional signs a, h, e, as in the case of Goshtasp.

[34]:

Of the last word consisting of nine signs he could read a . kh . a . .o.sh.o.h. i.e., all except the fourth and fifth.

[35]:

G did not have the signs for king at this point.

[36]:

Four words of which the first was identical with the last word in G.

[37]:

The same as the seventh sign in Goshtasp. The correct reading of the word for son was putra, identical with the Zend word.

[38]:

Ninety years later Grotef end's paper, because of its historic interest, was published in full by the Gottingen Academy, together with three others subsequently presented by him. See Nachrichten d. Kgl. Gesells. d. Wiss. 1893, No. 14, pp. 573-616.

[39]:

Ideen uber die Politik, den Verkehr und den Handel der vornehmsten Volker der alien Welt, 2d ed. (Gottingen, 1805). [English translation, London, 1854.]

[40]:

By A. de Caylus in his Recueil d'Antiquites Egyptiennes, etc., vol. v., (Paris, 1762), p. 79, seq., and Plate XXX.

[41]:

Observations sur les alphabets Zend et Pehlevi (Journal Asiatique, 1823, vol. ii, pp. 143-150) , followed by a more elaborate work in 1826.

[42]:

Commentaire sur le Yacna, (Paris, 1833) and Memoire sur deux Inscriptions Cuneiformes (Paris, 1836).

[43]:

Die Altpersischen Keilinschriften von Persepolis (Bonn, 1836).

[44]:

Instead of Kh.sh.e.h.i.o.h, it should be read Kh.sh.a.ia. t.i.ia the fifth sign being t and the sixth i, besides ia instead of h for the fourth and seventh signs.

[45]:

The papers of Hincks on the subject were published in the Transactions of the Royal Irish Academy for 1846-1847 ; those of Rawlinson in the Journal of the Royal Asiatic Society for 1846-1851; while Oppert's work on Das Lautsystem des Altpersischen appeared in Berlin in 1847.

[46]:

See A Memoir of Major-General Sir Henry Creswicke Rawlinson, (London, 1898) by his brother, George Rawlinson.

[47]:

Besides Persepolis, trilingual inscriptions showing the same three classes of cuneiform script had been discovered at Elvend, Hamadan, Murgab, Mesdjid, Mader-i-Suleiman, Naksh-i-Rustam, in addition to the large one on the rock at Behistun. See page 83.

[48]:

In 1907, Messrs. King and Thompson published the standard edition of Sculptures and Inscription of Darius the Great on the Rock of Behistun, in Persia, with the complete text of all three classes of the inscription, together with transliteration, translation and commentary. At the head of the inscription, the king has portrayed himself, surmounted by the symbol of his god, Ahura Mazda, in the act of receiving as prisoners a series of nine usurpers to the throne, whom he had succeeded in overthrowing.

[49]:

To make the method pursued by Rawlinson clear, let me remind the reader of the illustrations given above, p. 76, of names of persons which could be identified when a few of the letters comprising the name could be read. Similarly in the case of names of places. When, e.g., through having deciphered the names of Darius and Xerxes the signs for d, r, u and a had been ascertained, it was a comparatively simple matter to complete a name written with seven characters

. .u.d.r.a. . . .

by supplying m as the first letter and filling up the end of the line by ya and obtain mudraya which a Byzantine writer gave as the equivalent of Egypt and which, moreover, came close to the Arabic designation of Misr for Egypt. If, now, the same character to which Rawlinson had assigned the value m occurred in another proper name, he could readily decide whether the supposition was correct. In this way, as in working out a rebus, one conjecture was either confirmed or refuted by another.

[50]:

Thus, e.g., it was found that s in Sanskrit consistently changed to h in Zend and old Persian, Sindhus becoming Hindu in Zend, and Hindus, i.e., India, in old Persian; dasyu, "people" in Sanskrit, was ddhyu in Zend and old Persian, etc. It will easily be seen how important such a law of consonantal interchange was in interpreting endings to words, as well as words themselves in the Perse-politan inscriptions.

[51]:

Above, p. 82, note 45.

[52]:

Die Altpersischen Keilinschrift en im Grundtexte mit Uebersetzung, Grammatik und Glossar. A second edition appeared in 1881. Spiegel also embodied in his work a history of the decipherment.