^{page 240 note 1} I am indebted to Mr. A. G. Woodhead for reading and criticizing an earlier draft of this paper.

^{page 240 note 2} This period was chosen by me as the period of study for a doctoral thesis entided 'The Language of Attic Inscriptions, 323- 146 B.C. (excluding ostraka and vases)', sub mitted to the University of St. Andrews. Originally die chronological limits were set rather arbitrarily: the death of Alexander the Great provided a reasonable upper limit, but 146 B.C., apart from heralding the an nexation of Greece as a Roman province, did not—as far as my knowledge went at the time I began the study—have any particular linguistic virtue as a lower limit. However, it soon became apparent that after 146, or, more exactly, after the end of the second century, there appear phonological phenomena which occur only sporadically, if at all, before, and, on the other hand, some of the trends of the period I cover are reversed to give what the Greeks of that time no doubt considered the true Attic forms. This latter fact is particularly relevant to the phenomenon under consideration in this article.

^{page 240 note 3} All four combinations are considered in groups which consist of two elements. So here the possibilities are: , , and each occurrence of or counts as one. Thus counts as two examples of .

Footnotes 4 and 5 on the following page.

^{page 240 note 4} Each table is divided into 7 sections of roughly 25 years, so that the ratio of can be traced throughout the period. The total (T) is composed of figures taken from two dating categories (DC). DC 1 comprises examples drawn from inscriptions datable by archons, the date being based on the most recent archon-list put forward by Meritt, B. D. in The Athenian Year, pp. 231–8.Google Scholar It can now be claimed that the margin of error in dating those of the Hellenistic archons whose dates are still not absolutely definite is only a year or two. This has no significant effect on the present study. Figures quoted under DC 2 are drawn from inscriptions whose dates, although not completely cer tain, have a very good claim to reliability. These are (a) inscriptions dated from the occurrence of the name of a who is connected with the name of a known archon in another inscription, while at the same time the evidence of the type of docu ment, script, etc., point to the same date; (b) inscriptions in which a known archon is mentioned, but which do not contain the name of the current archon; (c) inscriptions dated with reference to a known historical event, and therefore inscribed at or near that date; (d) inscriptions dated from the men tion of a known .

The combined figure, therefore, should be regarded as giving a highly reliable indication of the number of instances.

^{page 240 note 5} These and subsequent figures for the period 323–146 B.C. result from an examina tion of all the relevant material in the follow ing sources: (i) Inscriptiones Graecae ii² (editio minor), 1913–40; (ii) Supplementum Epi-graphicum Graecum, vols, xii-xviii; (iii) He-speria, vols, i-xxx (including Supplements).

^{page 241 note 1} I do not think that the figures for the period 249–225 have any particular signifi cance, for two reasons, (i) In certain in scriptions is consistently chosen for a particular group (perhaps a reaction on the part of the drafter against the current trend), e.g. I.G. ii². 788. 7 (235/4); ibid. 18 and 22. (ii) If examples from less securely dated sources are admitted, the ratio of : is 25: 37. (Cf. also [d] below).

^{page 241 note 2} This additional evidence pertaining to the periods outwith the province of my thesis was culled from a cursory reading through the relevant inscriptions in I.G. and S.E.G., and should be interpreted as a fair indica tion of the situation rather tlian an exact analysis. It has, at least, the virtue of being a random sample.

^{page 241 note 3} It should be noted that, in estimating the number of examples of after 146 B.C., I have included those in which is written as . This tendency to omit the iota in , and , is very uncommon until the latter part of the second century, although examples do occur as early as the end of the fifth century (cf. Meisterhans, Grammatik der attischen Inschriften ³, p. 67). Towards the end of the second century there are several in scriptions in which the forms with iota stand side by side with those without it, e.g. I.G. ii². 1028 (101/100), where the editor notes: ‘Iota mutum in hoc titulo modo adscriptum, modo omissum est.’

^{page 242 note 1} Note the figures for the period 249–225 (cf. [a] above). For the reasons mentioned above I attach no special importance to these figures. Cf. I.G. ii². 790 (235/4), where the article appears three times with different words, each time in the form , viz. (line 11); (line 19). Again, if less securely dated evidence is accepted, the ratio of is 22:31.

^{page 245 note 1} I agree that the confusion was unlikely to occur. At the same time I am convinced that the only inference which can be drawn from the statistics is that the Athenians— whatever their motive may have been—did wish to make the distinction between the relative and the conditional particle quite clear. A question of conventional spelling is thus involved.

^{page 245 note 2} It should perhaps be pointed out that the reverse process, , gives only one example in the period under question, viz I.G. ii². 2498. 18 (321/0). Before 323 there are several examples, especially of , e.g. ibid. 17. 10 (394/3); 42. 5 (378/7), etc. After 146 I have noted only one example, viz. , ibid. 1046. 19 (52/1).

^{page 245 note 3} I am grateful to Professor K. J. Dover and to my colleagues in New England for their helpful criticisms of this section.

^{page 245 note 4} Certain prepositional phrases have a definite ‘formulaic’ flavour which tends to wards elision e.g. .

^{page 246 note 1} Cf. , I.G. ii². 463. 27 (307/6).

^{page 246 note 2} I.G. cautiously notes ‘notanda est hiatus cura’. Cf. the remark of Meisterhans, Grammatik der attischen Inschriften ³, sect. 24. 2: ‘Stärkere Elision, wie: erscheinen in Prosa nur verein- zelt’. This seems rather an overstatement, since I presume that ‘stärkere’ refers to the elision of and not of , and, if so, this example, which is the only one quoted by Meisterhans, is, as far as I know, unique and not ‘vereinzelt’.

^{page 246 note 3} I assume that the onus lies on the drafter and not the inscriber, since one would imagine that the Athenian authorities would engage the services of a reputable stone cutter, who would be expected to temper his poetic proclivities with a modicum of re straint.

^{page 246 note 4} ‘Frater Euryclidis eiusque potentiae et auctoritatis particeps' — Dittenberger, S.I.G.² i, no. 232. 34, n. 9.

^{page 246 note 5} I say ‘highly probable’ since there can be little doubt that the reference is to the quaestura militaris. However, the restoration should almost certainly be since, except in a rubric such as I.G. ii². 791, the article is added. Compare, for example, the formulae employed in the provision of a decree which indicates the official(s) responsible for allocating funds to defray the expenses of inscribing the stone, e.g. ibid. 844. 30–32, 66–67 (193/2); 950.22–23 (l65/4) etc.

^{page 246 note 6} To be dated 247/6 (not c. 232/1, as in I.G.)—see Meritt, , The Athenian Year, p. 234 (archonship of Diomedon).Google Scholar

^{page 246 note 7} The stemma familiae of Eurykleides can be found in Kirchner, Prosopographia Attica, no. 5966.

^{page 246 note 8} For the activities of Eurykleides and Mikion c. 217/6 see Polybius 5. 106. 7: … . About 213 B.C. they were put to death by Philip— (Pausanias 2. 9. 4).

^{page 246 note 9} This is rightly interpreted by Ditten berger as ‘primum locum inter eos qui de Athenis in libertatem vindicandis bene meruerunt, ii tenent qui civibus Piraeum reddiderunt, proximum post eos Euryclides et Micio’.

^{page 247 note 1} I do not mean to suggest that is a particularly poetic word-Professor Dover has pointed out to me that there are over 50 examples in the Corpus Demosthenicum, excluding Prooemia and Epistles-but that its forms are such as are adaptable to iambic metre.

^{page 247 note 2} Especially the latter two, since elision of is fairly common in Comedy.

^{page 247 note 3} It should be noted that it is tacitly im plied in my argument that is the last word of this iambic sequence. Accepting this, one can say either that the drafter con fined his poetic escapade to 5 feet or that he composed a full trimeter, the latter hypothesis necessitating a change in the restoration of line 17. Since, however, I cannot suggest any restoration which would satisfy both the sense and the metre, I feel that it is better to assume that initiates the iambic sequence.

^{page 247 note 4} The ‘I know thee who thou art’ con struction of the grammar books.

^{page 247 note 5} Professor Dover notes that the rarity of +subjunctive-rather than +subjunctive-in final clauses until the end of the third century B.C. suggests that it was chosen here for its metrical utility.

^{page 248 note 1} It should perhaps be added that throughout the surviving portion of this decree (a) where the ephelcystic nu can be used to avoid hiatus, the drafter uses it, e.g. lines 5 and 7; (b) in line 24, where he has his only other opportunities to elide the final syllables of a noun and a verb, he leaves diem in hiatus-. It is thus all the more remarkable tiiat he should feel so strongly about hiatus in line 18 if my interpretation is not correct.