settling the frequency of occurrence of individual words, though the statements "used only by Cicero" and "used first by Cicero," true for Greek works extant, might not have been true at the time of Cicero. Referring to the present mass of Greek, about fifty words occur only in Cicero, and a somewhat larger number are used first by him, though owing to textual uncertainties the exact number of each cannot be determined.

Exclusive of the quotations and proverbs, there are about 700 words - adverbs, adjectives, nouns, and verbs. In these, prefixal formations with ȧ-, ôus-, ev-, and prepositions are noticeable, and in the case of adjectives the number of verbals in -τός.

The paper was discussed by Professors Gudeman and Richardson.

14. Historical Note on Herodotus I. 106, by Professor H. C. Tolman, of Vanderbilt University (read, in the absence of the author, by Professor Fowler).

Herodotus (I. 95-106) gives the following events in the decline of the Assyrian Empire: (1) Median Revolt; (2) Revolt of the other subject tribes; (3) Conquest of these tribes by the Medes; (4) Median attack on Ninos (Nineveh), interrupted by the inroad of the Scythians; (5) Scythian supremacy (28 years); (6) Overthrow of the Scythians; (7) Fall of Nineveh at the hands of the Medes. The new stele of Nabû-na'id found at Hillah (Scheil, Recueil de Travaux, XVIII, 1896; Messerschmidt, Mittheilungen der Vorderasiatischen Gesellschaft, I., 1896) mentions: (1) Inroad upon Assyria on the part of the Umman-manda; (2) Devastation by these hordes of the temples of Assyria and destruction of the cities on the frontier of Accad; (3) Fall of Harran (date fixed by the inscription 607 B.C.); (4) Restoration of the temple of Sin at Harran by Nabû-na'-id in the third year of his reign (553 B.C.).

It is possible from this inscription to infer an alliance of the Babylonians and the Umman-manda. Both Berosus and Ctesias, although their accounts are widely divergent in other respects, yet agree on a Medo-Babylonian coalition against Assyria. If it be true that the Babylonians took a hand in the destruction of Nineveh, we can easily explain the omission of Herodotus on the ground that his informant was a Persian. However tempting the inference may be, yet the text of the document is far too mutilated to warrant Hommel's assertion that the Manda king coöperating with the Babylonian Nabopolassar razed Nineveh to the ground (cf. Billerbeck-Jeremias, “Der Untergang Ninevehs”). We fail to see in the inscription itself any direct reference to the fall of the Assyrian capital, and that too in the very place where we should most expect such reference.

Furthermore we cannot accept the theory that regards the Umman-manda as the same people as the Medes.

The Assyrian name (mât-Mada-a) generally given to Media (e.g. in inscriptions of Tiglath-Pileser, Sargon, Sennacherib, Esar-haddon) originally applied to an individual tribe, but later embraced all the scattered races. A beginning at least of Median unity is shown even in the time of Tiglath-Pileser (745-727) by his application of the epithet dannûti, "powerful" (Nimrud Inscription, 42). During the reign of Sargon (722-705) we find a Median confederacy so extensive

as to include several races not before classified as Medes (Sargon, 159 ff.). The inscription of Ashurbanipal (Cylinder B, col. III. 102-IV. 14) shows that the supremacy of Mata-a extended over more than seventy-five towns. If, according to the view of many Assyriologists, Mata-a be a variant for Mada-a, we can infer that the political union of Media had reached a high development in the last years of the reign of Ashurbanipal (668–626), a date which corresponds to that given by Herodotus for the beginnings of the Median dominion (646–624).

The old theory that these "united Medes" were designated on the later inscriptions Umman-Manda, "people of the Manda (Medes)," is without support. Tiele (Babylonisch-Assyrische Geschichte) regarded Umman-manda as an ethnological name for the "Medes of every race." Winckler, who has modified his old theory (Untersuch., p. 124 ff.), now holds that the term is strictly geographical, referring in a general way to the "tribes of the north," and, consequently capable of application to the Medes as well as to the Scythians, Cimmerians, etc. (Messerschmidt, p. 71 fg.). Delitzsch (Assyrisches Handwörterbuch, 1896) defined Umman-manda as “those northern hordes hostile to Assyria — i.e. Cimmerians, Mannaeans, Scythians, etc." As far as the etymology of the word can be determined it favors this view. The first member of the compound (ummân) signifies "people"; the second (manda) was connected by Jäger with ma-a-du, mandu, “much" (eg. Behistan Inscription, 20; Babylonian u-ku ma-a-du la-pa-ni-šu ip-ta-lah, Persian kárashim hacâ darshama atarsa, “the people feared him much"). But from the use of the word in the inscriptions we get our strongest evidence. It is clear that in the Sargon Annals (159 ff.) the context forbids any connection between Medes and Umman-manda.

Again, in the Behistan Inscription (II. 5) the Median Phraortes (Fravartish) claims descent from Cyaxares. Now if the Manda king Astyages had been the last legitimate Median king, there is no reason why the pretender should not have referred immediately to him. This, together with the fact that the Medes themselves gave over Astyages bound to Cyrus (Nabû-na'id-Cyrus Chronicle, Obv. col. II. 2), strongly favors the belief that Astyages, "king of the Ummanmanda,” šar amêl umman-manda (Nabû-nâ'id Cylinder of Abû-Habba, col. I. 32), was leader of those Scythian hordes which had overrun Media.

I believe it to be very probable that the Medes joined these northern peoples in the subjugation of Assyria. Such a union would not be without precedents. In the time of Esar-haddon (681–668) the inscriptions (Babylonian Chronicle, IV. 2) record an alliance of the Medes and Cimmerians against the power of Ashur. Furthermore, two hymns to the sun god (Sm. 2005; K. 2668) give the names of Median governors who coöperated with the northern invaders. But to declare (as many do declare) that we read on the stele of Nabû-na'id the fall of Nineveh at the hands of the Median hosts is not dealing fairly with our text.

If against the theory of Winckler and others we take the Umman-manda to be the Scythians (or even the Medo-Scythians), we have in this Nabû-na'id inscription supplementary evidence of that Scythian inroad mentioned in Herodotus (I. 106), an inroad which so weakened the Assyrian Empire as to make possible the fall of Nineveh; but since we have no information as to whether Nineveh fell before or after the devastation of Harran (607 B.C.), the exact date of its destruction must remain unsettled. In fact, we are still forced to admit that we possess no contemporaneous document describing this tremendous catastrophe.

AFTERNOON SESSION.

The Association assembled at 2.40 P.M.

15.

The Source of the so-called Achaean-Doric κový, by Professor Carl Darling Buck, of the University of Chicago.

It is an established fact in the history of the Greek dialects that the complete supremacy of the Attic Koh was for a time retarded by the spread in Western Greece, under the influence of the Aetolian and Achaean leagues, of another Koń, now commonly known as the Achaean-Doric Kový. So, for example, Brugmann, Griechische Grammatik,3 p. 22, after Meister and others. The thesis which this paper attempts to establish is that even this Kový is an indirect witness to the influence of the Attic Kový; for, although based in the main upon dialects of the Northwest Greek group, it is in a measure an artificial product, for which the Attic kon has furnished not only the suggestion but also certain specific elements.

Examples of Attic influence are: (1) the universal use of el for Northwest Greek and Doric al, (2) the use of πpŵτos in place of πрâτоs, (3) the prevalence of oi over Tol, (4) of iepós over iapós, (5) the frequency of eis beside èv cum acc.

To what may be called a second stratum of Attic forms belong also após in place of ποτί, εἶναι for εἶμεν, and forms like πόλεως, θάλαττα, τέτταρες, ἐάν, ἕως, etc. Aside from the question of Attic influence, the Aetolian кown and the Achaean Koh are to be distinguished in some features.

The paper appears in the American Journal of Philology, XXI. 193 ff.

16. The Sources of the Germania of Tacitus, by Professor A. Gudeman, of the University of Pennsylvania.

This paper appears in full in the TRANSACTIONS.

17. Pliny, Pausanias, and the Hermes of Praxiteles, by Professor Harold N. Fowler, of Western Reserve University.

This paper appears in full in the TRANSACTIONS.

18. An Inscribed Proto-Corinthian Lecythus, by Professor F. B. Tarbell, of the University of Chicago.

This vase is in the Museum of Fine Arts, in Boston, Mass.

The painted inscription reads:

Πύρ(ρ)ος μ' ἐποίεσεν ̓Αγασίλερο.

Pyrrhus, son of Agasileos, made me.

This is the only known piece of "Proto-Corinthian" pottery with a painted inscription. The date seems to be as early as the seventh century B.C., and the maker's signature is thus one of the earliest we have, perhaps the earliest. The character of the alphabet and the dialectic peculiarities point to Chalcis as the place of manufacture.

19. Note on a Certain Periodicity in Vital Statistics, by Professor Elmer Truesdell Merrill, of Wesleyan University.

The statement of the ages of the aggregate population of the United States, as reported in the eleventh census, shows what is seen also in the statistics of other modern nations, a tendency to the accumulation of ages at the even five-points, with an equally marked shrinkage at the unit-points immediately before and immediately after the even five-points. For example, taking the colored popula tion, there were reported in 1890, 23,000 persons of the age of 58, 16,000 of the age of 59, 78,000 of the age of 60, 13,000 of the age of 61, 15,000 of the age of 62, and so on. The extreme accumulation at the age of 60 is attained partly at the expense of all the surrounding ages, but more especially by the diminution of numbers at the ages of 59 and of 61. The average person anywhere between the ages of 55 and 65 may call himself 60, but he is more likely to raise himself to an even 60 when he has reached 59, and to keep himself at 60 when he has passed that age by only a single year. The same phenomenon is seen in the case of all classes of the population, and at all the even five-points, between the ages of 20 and 85. In the case of children and youth there is more precision of report: ages of 90 and above are not reported separately in the summary of the census.

The same phenomenon is seen also in the statistics of ancient Roman days. The nearest approach we have to census-returns is in the summary of Vespasian's census given by the elder Pliny (N. H. VIII. 153-164). Where he gives individual instances of great longevity he of course mentions them in multiples of five. But even where, as in the report of the 8th region of Italy, he seems to be giving the total number of persons a century old, or more, the statistics still run almost entirely by multiples of five.

More interesting is the study of the ages at death as given in the sepulchral inscriptions of the Corpus Inscriptionum Latinarum (cf. the tabulated statement prepared by Professor Harkness, in the Transactions of the American Philological Association for 1896, pp. 55 ff.). For each and every part of the Roman empire, and for every age from that of 20, or thereabouts, upward, precisely the same peculiarities of age-groups are seen as in our modern census reports. The only difference is that the exaggerations at and around the five-points are more strikingly marked in the Roman than in the American returns. Each of the volumes of the Corpus furnishes enough details to make a conclusion therefrom convincing, though, as might be expected, Vols. VI. and VIII. are richer in material than any of the others. As an example, in Vol. VI. the deaths recorded for the ages of 58 to 62 run in order, 12, 5, 91, 4, 15. The sudden fall immediately before the 60-point, the immense rise at 60, the corresponding drop at 61, and the recovery of the normal level at 62, after the fluctuation, are most striking, and are matched at the other decimal and semi-decimal points. The figures for the same years in Vol. VIII. are, 38, 12, 443, 107, 46. Here again the same phenomena are observed, and even in a more striking degree, with the exception that there is not such a great falling off at the 61-point as in the figures from Vol. VI. On the contrary, the number of deaths at 61 is disproportionately large, instead of small, and this same peculiarity is seen in many other of the multiples of five (plus one) in the returns from Africa. This must be added to the long list of peculiarities of the Province.

Not only is the exaggeration at the five-points observable in the totals, or in the case of those whose age at death is given in years only. It is seen as clearly where the age is specified up to the month or day. For example, these figures in Vol. VI. for the ages of 58 to 62 run, 3, 2, 11, 1, 4; in Vol. VIII. they run 6, 3, 10, 7, 5. There is a precise similarity, point by point, to the total figures for the same ages quoted before. Nor is the agreement strange. The precise birthday might well be remembered by the recurring round of its celebrations at the proper point in the course of seasons and festivals, even though the year was marked by no such unmistakable sign. The number of instances where the age is recorded to the very hour are too few to allow any deduction to be drawn from them.

The note was accompanied and illustrated by a series of charts prepared to show by a simple graphical method the total number of deaths reported at each given age in Vols. II., III., VI., and VIII. of the Corpus, and others to show that the same peculiarities are exhibited in the instances where the ages are reported more precisely than to the year alone.

20.

The Influence of Homer upon Tennyson, by Professor Wilfred P. Mustard, of Haverford College.

This paper is published in full in the American Journal of Philology, XXI. 143-153.

21. Some Affinities in the Maya Language, by Edmund Fritz Schreiner, of Chicago, Ill. (read by title).

The languages here compared are the Koptic, or language of the Egyptian people as we find it in manuscripts written between 250 and 450 A.D., and the Maya.

The latter is the idiom of about half a million of Indians in Central America. It is spoken in three distinct dialects or sister-languages: the Maya proper of Yucatan, the Quichè and the Cakchiquel, both in Guatemala. The Koptic words in the Comparative Vocabulary of the essay are taken from the Vocabu larium Coptico-latinum, which was carefully compiled by Dr. Parthey from the larger Koptic Dictionary of Peyron and from Tattam. It is claimed that Dr. Parthey's Vocabulary is thoroughly reliable, showing the phonetic form and the etymon in Latin of every word in each of the three distinct Koptic dialects; viz. the Sahidic of Thebes and Upper Egypt, the Memphitic of Memphis and Lower Egypt, and the Bashmuric of the Delta, the Oasis, and Fayum [wμe] around Lake Moeris. It was, therefore, possible to compare not only the Koptic in general, but also every Koptic dialect separately as to existing affinities; and it may be stated here that the investigation brought to light a very singular and interesting feature: Whenever words differ in the different Koptic dialects, be it in phonetic form or in signification, the Maya equivalent almost invariably agrees with the Sahidic dialect of Upper Egypt.

The Maya words for the schedule were taken from the Dictionario de la lengua Maya by Don Pio Perez, in which we find over twenty thousand Maya words with their Spanish etymon. This linguistic treasure has been built up by