^{page 67 note 1} S.H.A. p. 189. This example, as all the following, is as far aa possible taken from the most known and most accessible texts.

^{page 67 note 2} W.A.I. 4. 29. 58. In the published copy the last character has been left out.

^{page 67 note 3} W.A.I, 4. 17. 22.

^{page 67 note 4} S.B.A. vol. iii. p. 536.

^{page 67 note 5} Pinches, T., “Texts,” p. 4, 1. 7Google Scholar.

^{page 67 note 6} S.H.A. p. 189.

^{page 67 note 7} Ibid. p. 108.

^{page 67 note 8} Ibid. p. 72.

^{page 67 note 9} Ibid. p. 189.

^{page 68 note 1} S.B.A. vol. v. p. 107.

^{page 68 note 2} Ibid. vol. iii. p. 567.

^{page 68 note 3} S.H.A. p. 225.

^{page 68 note 4} Ibid.

^{page 68 note 5} Tablet of Gram. Forms.

^{page 68 note 6} S.H.A. p. 257, see also p. 251.

^{page 68 note 7} Ibid. p. 180.

^{page 68 note 8} W.A.I. 4. 17. 40.

^{page 68 note 9} S.H.A. p.

^{page 68 note 10} W.A.I. 4. 29. 48.

^{page 68 note 11} S.H.S. p. 142.

^{page 68 note 12} W.A.I. 4. 9. 58–60. 1–10. All these examples are by no means exceptional; it would have been easy to multiply them, but these will be sufficient to illustrate the actual paper, as the object is not to give a complete survey of the pronouns, but merely notes.

^{page 69 note 1} Assyrian Grammar for Comparative Purposes, 1872, p. 39.

^{page 69 note 2} S.H.A.

^{page 69 note 3} Given by Prof. Sayce, but I have not found any example in the texts yet examined by me.

^{page 69 note 4} W.A.I. 68, 1. 59.

^{page 70 note 1} W.A.I. 29. 26–34.

^{page 70 note 2} The examples for the first person are very numerous in the historical inscriptions. See S.H.A. and S.H.S.

^{page 70 note 3} W.A.I. 4. 14. No. 2, et seq. See also W.A.I. 4. 9. 54 et seq., etc., and S.H.A. for the 1st person.

^{page 70 note 4} Sometimes the long vowel is not expressed: šasi, šašu, šaša.

^{page 70 note 5} Pinches, T., “Texts,” pl. 4, 1. 8Google Scholar.

^{page 71 note 1} We found also ša-šunu ‘them,’ the forms appear to be oblique cases, being really the relative in regimen with the pronominal suffixes; many examples are found in the Deluge Tablet, S.B.A. vol. iii. pt. 2. Smith took them for oblique cases of personal pronoun šū (ibid. p. 589).

^{page 71 note 2} The late S. Guyard had already assimilated this suffix to the adverbial formative iš in Assyrian.

^{page 72 note 1} S.B.A. vol. vii. pt. 3, p. 382.

^{page 72 note 2} Flad, M., A Short Description of the Falasha and Kamants, St. Chrishona, 1866Google Scholar; Halévy, J., Essai sur la langue Agaou, Paris, 1873Google Scholar.

^{page 72 note 3} Waldmeier, Th., Wörter-Sammlung aus der Agau-Spraehe, St. Chrishona 1868Google Scholar.

^{page 72 note 4} Reiniṣch, Leo, Die Chamir-Sprache in Abessinien, Wien, 1884Google Scholar.

^{page 72 note 5} Reinisch, Leo, Die Bilīu-Sprache, Wien, 1882Google Scholar.

^{page 73 note 1} Almkvist, H., Die Bischari-Sprache, Upsala, 1871Google Scholar. The Bishari is spoken in Southern Egypt between the Nile and the Red Sea.

^{page 73 note 2} Reinisch, L., Die Kunama-Sprache, Wien, 1881Google Scholar. The Kunama is spoken at the West of Abyssinia.

^{page 73 note 3} In Agau, and all this group of languages, the possessive pronoun is prefixed to the object possessed: Agau yi-nañ= Assyrian biti-ia ‘my house.’

^{page 73 note 4} ana is given by Flad, yin given by M. Halévy is the Hamara form, which may have passed into Falasha.

^{page 73 note 6} Flad, op. cit. p. 29.

^{page 74 note 1} Küten is formed as kašunu and šuatunu in Assyrian by adding the mark of the plural n to the singular after the suffix.

^{page 74 note 2} The plural common is yūk, kūk, etc., being formed by one of the plural formative -k.

^{page 74 note 3} The reduplicative form šūšū, being used for ‘one another,’ gives the primitive meaning of ‘one’ or ‘he.’ In Bilen šūtšūt or, with assimilation, šuššut is the reciprocal pronoun.

^{page 74 note 4} The χ answers to the Arabic

^{page 75 note 1} Between those two modes of expression there is exactly the same difference in Assyrian and in Agan. The postfix ti marks a kind of conditional of consequence, but the postfix -ni in Assyrian, -na in Hamara, marks a conditional of simple sequence or simply indicates that the clause is relative. Strange to say, this postposition is found in active use only in Galla, where it serves to express the ablative.

^{page 75 note 2} Ch. Tutschek, , Grammar of the Galla Language, Munich, 1845Google Scholar; Massaja, , Lectiones grammaticales, Paris, 1867Google Scholar.

^{page 76 note 1} The apparently irregular formation of the Galla pronouns might be explained, but it would carry us too far, and for this reason I have not given all the forms, which would have necessitated more explanations. The emphatic pronouns anatŭ, zitŭ, etc., have much puzzled Tutschek, who could not explain them but by supposing a misuse of the dative, the final of which had been weakened; these forms, however, are easily explained when compared to those of Hamara.

^{page 76 note 2} Reinisch, Leo, Die Kunama-Sprache, Wien, 1881Google Scholar. The author of this grammar had first classed Kunama with Nubian, but since he has replaced it in the Kushite group.

^{page 76 note 3} Almkvist, H., Die Bischari-Sprache, Upsala, 1881Google Scholar. This language is interesting for the large development it has given to the distinction of gender, and in that it goes even further than Berber.

^{page 77 note 1} I was inclined at first to see in the enclitic -sa a remnant of the dative suffix, but this enclitic is used exactly in the same way as the Assyrian enclitic -ma, and we must see in it no doubt the Agau šū ‘self.’ What is found in Amharic and Tigré, connected to Agau, must be of modern introduction, as it is not found in Ethiopian, Himyaritic, or Arabic.

^{page 77 note 2} Elements of Comparative Philology, London. 1862, p. 602Google Scholar. In this work the author practically classes the Semitic dialects with the African tongues, bringing them in close connection with the the Agau group.

^{page 77 note 3} The grammars of Dr. Reinisch are really the first scientific works on these tongues, without which we would be unable to make any comparison. The notice of Flad is too short, and is really a mere collection of notices without any explanation; the notice of M. Halévy is too unscientific to afford trustworthy materials, the forms given in his extracts are not even explained, and it would be impossible to get an idea of the structure of the Falasha dialect from his notice alone.

^{page 78 note 1} I do not maintain that every language has passed through such a stage, but Egyptian certainly has; in the inscriptions of the first dynasties it is often difficult to detect what is the person meant, the context and the ideographic system of writing is the only guide. The first person and the second are often expressed by the same suffixes. It is only after many trials, and a long period of distribution, that Egyptian arrived at something like a determined paradigm, but the incertitude and hesitation have left traces even in the Coptic. In the Semitic tongues a remnant of this period is found in the suffixes of the aorist past, which distinguishes the first and second person only by the vowel.

^{page 79 note 1} The promiscuous use of these themes explains why, in the Turanian tongues, there is often a different localization for the use of such or such a theme; it is easy to see in examining the tables compiled by Lenormant, M. (La langue primitive de la Chaldée et tes idiomes Touraniens, Paris, 1875)Google Scholar.

^{page 79 note 2} In these remarks I will not give any illustration, as the texts in W.A.I, are easily accessible; but those who are not familiar with the Cuneiform writing I refer to the works of M. Lenormant, who gave in his Etudes Akkadiennes numerous texts with transcriptions and translations.

^{page 79 note 3} The distinctions often escape us, and were also no doubt rather loosely indicated by the Akkadian, as in our modern tongue the difference of ‘this’ and ‘that’ is not always accurately noticed; variants give ab for an in the texts.

^{page 79 note 4} In that case the context, or the presence of a particle or postposition without regimen, can only denote the existence of the theme which has disappeared.

^{page 80 note 1} W.A.I. 5. 22. 67 et seq. me-a me-en is translated by ia'nu anaku and ia'nuatta, W.A.I. 5. 40. 6–7, etc.

^{page 80 note 2} W.A.I. 5. 27. 34, and also found in a text.

^{page 80 note 3} It is rather too early to explain or seek the origin of these suffixes; however, the use of mulu or lu ‘man,’ for the first person, might lead us to give to mu, which is sometimes translated by the Assyrian zikaru ‘male,’ a similar meaning, zu might be compared to su ‘body,’ the second person being considered as more passive. It must be said, however, for that some scholars give it the reading ku in the Akkadian texts, and mu only in the dialectical texts; if this is so, we would have the theme ku given also for the first person (W.A.I. 5. 20. 57) and the second (W.A.I. 5. 27. 35). and it might be the same as the disused article preserved in a few words (S.B.A. Proceedings, Nov. 1882).

^{page 80 note 4} is read mal by some, but the expressions ma-da ‘to me’ and ” and ma-ra ‘towards me,’ and the plural me, seem to establish the reading ma. It may be however pronounced according to the dialects ga or ma, as is supposed to be pronounced gu and mu.

^{page 81 note 1} As it will be seen, the distinctions made by the scribe are not always confirmed by the texts.

^{page 81 note 2} I pass over this part, which has little bearing on pronouns.

^{page 82 note 1} I place between brackets the translation of the ideograms , etc.

^{page 82 note 2} l. 30 ka-nam (supposing this reading correct) would be ‘to the face,’ which answers well to aššum ‘as for.’

^{page 83 note 1} Journ. R.A.S. Vol. XVI. p. 318.

^{page 83 note 2} It would be wrong, I believe, to confuse or assimilate this pronoun with that of the first person.

^{page 83 ntoe 3} This first ab seems to have been written in error, and maintained to avoid an erasure.

^{page 83 note 4} Opposite this last form seems to be anaku ditto, but this Assyrian expression is rather placed between 11. 24 and 25, and refers no doubt to the next series.

^{page 85 ntoe 1} If the reading of this character is ma, it is no doubt a form for me-a, where the vowels have been assimilated, but it would in pronunciation be confused with the singular of the pronoun, it may be to create a graphic distinction that the scribe has written three wedges inside instead of two. In the group and the first character would be ma for me-a, a kind of accusative followed by the postposition.

^{page 85 note 2} This pronoun is remarkably similar to the interrogative aba ‘who.’

^{page 85 ntoe 3} In the texts we found zae and zā used with a postposition.

^{page 86 note 1} W.A.I. 4. 21. 3–4.

^{page 86 note 2} Kunu-ma is not found in the texts, but is formed regularly as iatima and attama.

^{page 86 note 3} W.A.I. 5. 40. 5.

^{page 86 note 4} Za-a is found in the texts.

^{page 86 note 5} which is given as synonym of ninā and na, may be for ε-šu ‘to him,’ transformed into eši by assimilation, and then the theme e was dropped.

^{page 87 note 1} There is no doubt that in the Assyrian copy ana is for anaku, it may be the fault of the copyist, or perhaps an Aramæan influence.

^{page 87 note 2} This Assyrian expression is followed by the sign ti, the meaning of which escapes me, it might be intended to express that this combination is used in incidental sentences, as we hare seen the postfix ti indicates sometimes an incidental or secondary clause.

^{page 87 note 3} What prores that this distinction is only indiridual is that these very same forms appear in the texts without haring these distinctions.

^{page 88 note 1} After line 28, the scribe gives for each series only the first person in the translation, leaving the rest to the reader's intelligence.

^{page 88 note 2} This had already been perceived by M. Lenormant.

^{page 88 note 3} I suppose that is short for šuati, the two small wedges representing the syllable which is not repeated.