^{page 135 note 2} Apart from these, there is a small group of characters whose origin is older than the papyri. These in some cases are epigraphic forms retained, but others may have had a manuscript origin.

^{page 136 note 1} It is surprising that hundreds of common words have not been forced into abbreviated form in modern English. As for perspicuity, who could possibly mistake the meanings of: wh., c^d, w^d, etc., in any context? As regards economy the mere saving in type, time, and space would be enormous.

^{page 136 note 2} Cp. the habit of Mediaeval writers of Tironian notes, who supplied signs for word-endings where the ancients had left the notae abbreviated.

^{page 137 note 1} Concerning the nature of the abbreviation used in the Ἀβηναίαν Πολιτϵία of the British Museum, and two or three others like it, which stand apart from the ordinary literary papyri written in careful uncials, judgment may be at this stage wisely suspended. See Gitlbauer, Tachygraphische Spuren im Papyrus der aristotelischen Πβλιτίία, Ἀθηναίων, in the Archiv für Stenographie July-Dec. 1901Google Scholar. In this learned contribution the ‘special pleading’ for formal tachygraphy is undisguised. I attempt a less ex parte estimate in a forthcoming contribution. See note at end, p. 173.

^{page 137 note 2} For a doubtful exception, vide infra p. 148.

^{page 137 note 3} And also in Attic inscriptions from the fifth century B.C. onwards.

^{page 138 note 1} In Pet. Pap. XXXIII (a) 24 early 3rd cent. B.C.

^{page 138 note 2} In B.M. Pap. CCCXCIII, 3 (saec. VI–VII). So the ‘10,000’-symbol which occurs among the Petrie-symbols, occurs more than 8 centuries later in Oxyr. Pap. CXXVII recto (late sixth).

^{page 140 note 1} Wilcken mentions it, in his Dissertatio adsummos honores written in 1885. Dr. Kenyon gave it to me last year as the accepted opinion, so that it has survived a long period of criticism.

^{page 140 note 2} Once in this papyrus followed closely by με = μεγάλη(?): apparently the metropolis, Crocodilopolis (Mahaffy II p. (87). M. elsewhere quotes εκ κροκοδιλων ʘ from the heading of an account dated 20th year of Ptolemy III (226. B.C.).

^{page 140 note 3} Cp. Fay. Pap. XVII. 1. Here, however, it is further distinguished by the ordinary horizontal inclusion-sign.

^{page 141 note 1} The Revenue Papyrus, of this century, has α/,αρ, and αρ as common abbreviations for Artabas. The Petrie Papyri have a conventional form

^{page 141 note 2} This horizontal-mark, which is found in the oldest manuscripts and onwards, probably indicates originally not so much the omission of the absent letters as the inclusion of all the letters which it covers in a compound with a special meaning. This would be necessitated by the habit of continuous writing, without division of words, etc. The reader is warned by an over-written horizontal to look for some special meaning in the included letters, which otherwise being taken in conjunction with the letters of the context might accidentally make new meanings with them. This sign contributes to the formation of a system of expressing fractions, which is commonly employed, but it would seem to be of general rather than particular use, as fractions are sometimes expressed in quite another way, viz. by drawing a vertical stroke to stand directly over each member. In B.M. CCXXIII. (second cent, B.C.) we have this marking for one-eighth, one-sixteenth.and one-thirty. second, the one-eighth again in lines 6, 13, while (a less certain reading) CCCCII. 5 has other fractions which appear as vertical strokes with a formless thickening at top and bottom. It is seen again in the common ‘one-quarter’ symbol (p. 147 infra) and probably, reduced in size, in the Roman forms of 4-chalci (CCCCLX. (A.D. 191) lines 2, 3, 4, 5, CCCXII. (A.D. 147) line 7), ‘one per cent, and two per cent.’ (CCCVII. 2).

^{page 142 note 1} Cp. also Wilcken, , Gr. Ostr. i. p. 819Google Scholar for (Alpha surmounted by Upsilon. W.'s own explanation) in the ostraca. Revillout, , Lettres sur les monnaies Egyptiennes, 1895Google Scholar, gives (pp. 172, 3) a slightly different conventional form which he thinks to be Alpha-Rho, with the over-written bar. It occurs naturally in Ptolemaic cursive. The somewhat injured example B.M. XV. (6), 5 is most like W.'s.

^{page 142 note 2} In one place at least, Pap. XV, fr. 8, line 2 αν̆ is used not meaning ἀνά but something like the σιτώνιον or ὀψώνιον of items in the context since it is worth τι dr., a meaning necessary to make the total given. But note that this is (1) not the common form of the superscript Nu; (2) not the ordinary ἀνά of the same MS.e.g. line 1 Variants in fr. 13: CCCCII. r. line 5.

^{page 143 note 1} Mahafly explains a repeated occurrence, all down a column, of the fully written word ἀνά as equivalent to our × (‘multiplied by’) thus: ιε ανα δ∠ ξζ ∠ where the arithmetic (15 × 4½ 67½) supports that meaning. Vide fac. of Petrie Pap. II, XXX. Of course this is almost the same thing as ‘at the rate of.’

^{page 143 note 2} Wilcken cites and illustrates a number in the second part of. his Observationes ad historiam Aegypti…depromptae e papyris Graeeis Berolinensibus ineditis, 1885. But the instances seem to be drawn from Paris Papyri.

^{page 143 note 3} That is to say, for instance, the scribe of Pap. Par. no. 66 in writing or the scribe of (for Gamma circumflexed) is not writing these peculiar forms as he would write the same letters in the context, even if he afterwards intended to add a circumflex.

^{page 144 note 1} The Kappa-shaped symbol, with the horizontal bar, is, as Mr. Hill reminds me, a not uncommon monogram of κε (=και) in Imperial times.

^{page 145 note 1} Vide King & Cooksoi), Compar. Gram, of Gk. and Lat. p. 26Google Scholar.

^{page 145 note 2} Dr. Kenyon suggests that the former of these forms may be simply an arbitrary variant of Mr. Hill suggests that the latter is derived from the sign used for σσ in some early alphabets; and that, if this sign followed ω in the alphabetical sequence, it would naturally be used to represent 900. As this completes the alphabetic representation of all the numerals, it seems to me extremely probable.

^{page 145 note 3} B. M. Pap. XXVIII, 8 has which is edited as 200 dr.: this is however of the middle of the second cent. B.C.

^{page 146 note 1} Third cent, (B.C.) forms Pet. Pap. XXXIII (a), 24, and 32, which show both. p second B. C. and later.

^{page 146 note 2} Mahaffy in the Introduction to the second part of the Pet. Pap. gives this list for obols: and remarks on the ¼-obol symbol: “(It) is so various that I am not sure whether it only represents one fraction.” For the ½-drachma symbol, Dr. Grenfell has the same shape in the index of the Revenue Papyrus.

^{page 147 note 1} In the same work at pp. 172–3, K. incidentally gives a fresh illustration in artabe.

^{page 147 note 2} It is worthy of note that Dr. Grenfell gives in his index for the Revenue Papyri the form ∨ for this symbol.

^{page 148 note 1} In The fully formed Beta is written in CLXXI (a) 6, a Roman (102 A.D.) tax-receipt.

^{page 148 note 2} The appearance of this symbol is very varied. In Ptolemaic papp. it is often very like the Talent-symbol, with which Forshall the early Brit. Mus. Editor actually confuses it in text of XV (13) reverse, and on reverse of XV (14) with ∽̆ Aroura (Ken. ad loc).

^{page 148 note 3} Wessely, Éin System altgriechischer Tachygraphie Taf. 1; or Rainer Pap. Taf. XIII. Nr. 444, a tachygraphic papyrus of the V–VI cent.

^{page 148 note 4} After this it is common in later Greek taehygraphy.

^{page 149 note 1} Die drei Systeme der Griechischen Tachygraphie, Taf. 11.

^{page 149 note 2} I do not consider that the peculiar variant which Dr. Mahaffy notes as occurring all through Pet. Pap. XXXIII (vide Part II, 35) is important against it, as he points it out as exceptional, in the third cent. B.C. Of course, for what it is worth, it does favour the Alpha theory.

^{page 149 note 3} There is an example in a Roman (2nd cent.) pap. B.M. CCCXXXIX, 15, which, accidentally, illustrates a possible Epsilon-stage. With such a form the development might have been (or ligatured )— or or

^{page 150 note 1} He quotes J.H.S. xix p. 15, no. 9 to prove its occurrence in lapidary inscriptions.

^{page 151 note 1} It is perhaps worth noting that the Phoenician letter Heth actually became used as the sinn of rough breathing, in Alexandrian grammars, and, at an earlier period, in some epigraphic alphabets; while, curiously enough, a fuller form (? of the drachme-symbol) viz. is noted (Mahaffy Part II pp. 39–41 on Pet. Pap. VIII (1)) as being found ‘where we should expect drachme,’ Lastly, to add another ebment of perplexity, Plate XLVI in Part II of the Pet. Papp, has in line 15, the symbols which Mahaffy translates 325⅚.

^{page 151 note 2} Subsequently, Dr. Kenyon writes: ‘must indicate the object for which the money (10 dr. in 1. 2, 85 in 1. 8) is paid.’ This value for the sum of drachmae removes the chief difficulty, and increases the probability that-the unknown symbol is simply =αλ= ‘item’ ‘ditto.’

^{page 152 note 1} E.g. in the case of vide Kenyon B.M. Pap. Cat, p. 56.

^{page 153 note 1} The expedient for instance is largely used in modern reporting, e.g. in ‘phonography,’ where e.g. n crossed by t has the meaning ‘notwithstanding’; tmp crossed by s becomes ‘temperance society.’

^{page 153 note 2} The Ptolemaic ⊢= ‘drachmae’ does not survive; the Ptol. or now appears in the last shape, with the horizontal lowered and bent generally.

^{page 154 note 1} The sweeping curve of the Roman and By zantine periods, a semicircle with its convex side to the right, is not also Ptolemaic.

^{page 154 note 2} Wessely, (Ein. Syst. p. 8Google Scholar) mentions metrological sigla, figures and fractions, in cursive texts of Berlin and St. Petersburg. But he gives no details.

^{page 154 note 3} καί has also been suggested as a meaning (vide Brit. Mus. Pap. Cat. Vol. II. Index of Symb.).

^{page 154 note 4} An interesting and rather extreme example of this use is and in CCCXV 18. Cp. variants CCCXXIX 23 (rising high above the line). Cp. also Oxyr. Papp. CCLXXXIX (1) 12, 19; CCXC, 20, 23.

^{page 154 note 5} Only incidentally does it mean ἡμιὠβολον (so Edd. of Fay. Pap. pp. 181 and 347). Vide supra, pp. 146 sqq.

^{page 155 note 1} That is, the printer's types of these shapes.

^{page 155 note 2} Cp. the same stroke again in an unusual abbreviated ῾Ηφαίστου viz. Ηφ∫̅ CXXXI r. 329, 372. Cp. CCCXXV (a) and (b) where (a) line 5 has the sign for ‘year’ absolutely indistinguishable from the sign of abbreviation in (b) line 1.

^{page 156 note 1} The same demand for definiteness in later days is seen in mediaeval commentaries of Notae Tironianae. In these we get careful restorations of case-signs and geneial word-endings where the earlier notarli had been satisfied to trust to memory and the context. The in creased exactness is in all these cases in inverse ratio to freshness and spontaneity, and easy familiarity with the language to be written.

^{page 156 note 2} See his interesting study of the symbol in Archin für Stenographie, Berlin, January, 1902.

^{page 156 note 3} CXIX, 4.

^{page 157 note 1} Χάρακος vide p. 158 infra.

^{page 157 note 2} CXIX, 5, cp. = λαχ (ανία)

^{page 157 note 3} CCLXXXVIII, 4 et saep.; CCLXXXIX, i, 10, et saep.

^{page 157 note 4} Grenfell aud Hunt print it (Fay. Pap. XI, 15, p. 347, etc.) as a plain right angle.

^{page 157 note 5} Which in Ptolemaic papyri has a wellmarked rounded variant.

^{page 157 note 6} These are not new in the Roman papyri and have been explained under Ptolemaic (p. 146 sqq. supra).

^{page 157 note 7} In a few places there is used a method of expressing numerator and denominator. See par. on fractions inf. p. 160 sqq.

^{page 158 note 1} Privately communicated, May 1901.

^{page 158 note 2} That is to say, in this particular context. For a Chi=choenices certainly occurs, cp. Oxyrh. Pap. CCLXXXVII, 7, 8.

^{page 158 note 3} A slip, I think, for

^{page 158 note 4} Its variants boldly formed however occur early. Cp. CCII, late 1st. cent.

^{page 158 note 5} The common Ptolemaic ⊢ is not Roman.

^{page 160 note 1} CCCCLXXVIII, 6 offers an exception.

^{page 160 note 2} Diophanti Alex, opera, ed. Tannery, P., and Hultsch's, review of it in Berl. Wochenschr. 1894 p. 805Google Scholar.

^{page 161 note 1} The Edd. of the Fay. Pap. LVI, 5, 6 LVII, 5, resolve into χαλκου̑ ὀβολοί Perhaps the explanation is that when used alone, the collocation is to be read in this way, but that when following ό = 4 chalci as B.M. Papp. CCCXII and CCCLXXI, it adds 2 chalci, making up the 6 chalci.

^{page 161 note 2} We have in Roman papyri the three common variants: (1) the plain right angle resting on one side of it; (2) the same with its vertical member curved, 2-shaped; (3) the same again, but with the concave of the curve looking to the right. Both the (2). and (3) may be seen in CCLVII A.D. 94 and CCCXXIV A.D. 161. For comparison of the last with its cursive degenerates CCCXIV 25, 26. Grenfell and Hunt print other variants ∠ (Oxy. Pap. Vol. I. p. 263) and ∠ (ib. vol. II. p. 337).

^{page 163 note 1} In CXCII col. 4, line 82 it apparently is used for ἀρταβω̑ν

^{page 163 note 2} ὀρόβου a kind of vetch or pulse.

^{page 163 note 3} The πυρου̑ ordinarily not being represented among the symbols.

^{page 163 note 4} This is dated 150 A.D. so that the use here is an anticipation of what became general later.

^{page 164 note 1} Cp. XV (8) 9 which is indexed in the B.M. Cat. (Kenyon, 1893) as above, with the ‘metretes’-symbol in the same index.

^{page 164 note 2} The latter would be more in accordance with the regular principle of abbreviation, but the apparent occurrences of Omikron make the former worthy of consideration.

^{page 164 note 3} A curious Ptolemaic example is worth note in this connexion, viz. in CCXVII, 4. Here the Pi-Omikron composite is already crossed by a horizontal midway. This cannot represent the ‘artaba,’ for which a separate symbol is written. The date is third cent. B.C. (?) or latest second cent. It must be noted that = πυρου̑, and = πυρου̑ ἀρτάβαι is found also in the Ptolemaic period, on the ostraka, e.g. B. M. No. 25868.

^{page 165 note 1} The Omikron circlet attached on the under side to the right hand end of the line, reappears, and survives to Byzantine papyri. Cp. in the Abinnaeus papyri, CCCCXXVIII nearly 40 occurrences, CCXLIX, 20, CCXXX.VI, 4. In CCXVII, 16, (3rd cent.) it is placed above the horizontal.

^{page 167 note 1} For the latter, see Fay. Pap. XXI, I (a) 9, 10, LXXXVII, 1, 10,13.

^{page 167 note 2} Sir E. M. Thompson remarks that the πνεύματα are not found in early Gk. MSS. before the 7th cent, and did not become rounded until the 12th cent. Gk. and Lat. Pal. pp. 71, 72).

^{page 167 note 3} Exceptions are to be seen in the rare arrangement e.g. = δεκατάρχης (Oxy. Pap. LXIV, 1); and χ= ἐκατοντάρχησ = ἑκατοντάρχης (ibid. LXII, 1); the numeral has the middle letter of tho αρχης written over it. Cp. Genev. Pap. No. 35, etc.

^{page 168 note 1} CXXI, 927 bas a cup-shaped dot which may be part of a conscious Omikron.

^{page 170 note 1} It is used apparently with a plain numeral CCXLIX, 32.

^{page 171 note 1} CCCCXXVIII for instance, oue of the Abinnaeus papyri, has not the dots (8, 22, 23).

^{page 172 note 1} And may be due only to a kind of attraction to such forms with ordinary ligatures as in CCXLIX, 20.

^{page 173 note 1} In the May number 1902. Dr. Kenyon, the first editor of the papyrus, has kindly read the manuscript of this article, and expresses his agreement with the general conclusion thereof.

^{page 173 note 2} Apart of course, from pure continnous tachygraphy.