Alphabetization in Harpocration's "Lexicon" Keaney, John J Greek, Roman and Byzantine Studies; Winter 1973; 14, 4; ProQuest pg. 415 Alphabetization in Harpocration's Lexicon John J. Keaney I N HIS VALUABLE Contributions,! Professor Daly has shown that the principle of first-letter alphabetization in arranging lists only became systematically used in Alexandrian times, and it is only in the second century that there is a tendency toward the use of absolute alphabetization: the two extant works which show the latter are Ga­ len's Interpretation of Hippocratic Glosses and Harpocration's Lexicon of the Ten Orators. Daly thought that the alphabetized order of the first was imposed by the author,2 and that the order of the second is due to a later, Byzantine revision. Since KUhn's text is of little value in this area, the truth about Galen will have to await the new edition in the Corpus Medicorum Graecorum. It is, however, interesting to note that the two scholars who worked most closely with this text earlier were of contrary opinion about the degree of its alphabetization: Ilberg3 suspected that the true order was represented by the first-letter alphabetization of Marcianus Gr. 269, while Helmreich4 used the principle of more rigid alphabetization to make a series of correct (and verifiably so) emendations. Doubtless Helmreich's assumption will turn out to be closer to the truth. For Harpocration the situation is somewhat more complicated than is reflected in Daly's discussion. It is clear that the author of the Lexicon adopted (though not slavishly) absolute alphabetization as his major principle of organization: it is also clear that there are series of lemmata in which this principle was not fully carried out and clear as well that the alphabetic order was disturbed in the course of the 1 L. W. Daly, Contributions to a History of Alphabetization in Antiquity and the Middle Ages (Collection Latomus 90, Brussels 1967) esp. 32-35. 2 Galen's use of this principle was in response to a request: ~CTat 8i, wc a&r6c ~Ki>'€vcac, Tj Ta.,tC TCP >'6yCfI KaT6: n}v TWV ypall-ll-d.TWV Ta.~tV (19.63 Kuhn). A similar notion (OJc TtV€C ~(N)'ovct) is found in the introduction to the pseudO-Galenic (alphabetized) II€p~ aVTtfJaMoll-ivwv (19.723 Kuhn) 3 }. Ilberg, De Galeni vocum Hippocraticarum glossario (Leipzig 1888) 340. 4 G. Helmreich, "Handschriftliche Verbesserungen zu dem Hippokratesglossar des Galen," SB Berlin 1916, 197-214. 415 416 ALPHABETIZATION IN HARPOCRA TION'S LEXICON transmission of the text. Since Harpocration was among the first to attempt complete alphabetization, a relatively full discussion of the problems he faced and the results he achieved may be of interest.S I Our knowledge of the text of the Lexicon, which can now be firmly dated to the second century of the Christian era,6 is based on an epitome (= Ep) made sometime before A.D. 850 (although the MSS are much later) and on a fuller recension (= ll), none of the MSS of which is earlier than A.D. 1300. There is also one small papyrus fragment (P.Ryl. 532= Pack 2 458), which is of major importance for the light it throws on the alphabetization of the text. In format, the Lexicon is a list of words or phrases (lemmata) found in the Attic orators: each gloss begins with the lemma; an explana­ tion of varying length follows, together with a reference to the orator(s) in which the lemma is found. Clearly an alphabetic arrange­ ment is the most convenient one for this format, and, except for two glosses in which the principle is discarded, no other arrangement is used. That is, the lemmata are not restricted to anyone speech or to anyone orator. Nor do they fall into one category: included are legal terms, cult language, personal names, place-names, etc. The Lexicon contains 1247 glosses, of which slightly less than ten per cent break the alphabetic order.7 This percentage, itself perhaps com- 6 Daly's suspicions of a later rearrangement were based upon a general assumption that rigid alphabetization is a sign of late revision and on particular observations (p.32): "The suspicion of re·arrangement is strengthened to some extent by the fact that the upsilon of Tavp€ac is alphabetized as beta, the iota of ,un] as eta, and nasalized gamma as nu. Another feature of the arrangement of this work is that a simple word like 8£cp.Oc precedes its compounds such as 8£cp.o8€'Tal. ... The general assumption (cf. e.g. J. Tolkiehn, RE 12 [1925] 2434) was formulated and fostered before the discovery of papyrus lexica. While it is true that only one of these of any length (P.Oxy. 1802, saee. II/III p., 20 items) shows absolute alphabetization, this suffices to show that the assumption cannot be automatically applied. There is in fact no sign of this kind of revision in Harpocration, nor is it easy to see at what period it could have taken place. The single papyrus shows the same order as the MSS, the Lexicon was alphabetized before the epitome was made, and nothing at all is known of the work between the date of the papyrus and the ninth century. I discuss Daly's particular observations in the text (with note 12) and only note here that ATfT11 is the correct form. e Cf. E. G. Turner,lEA 38 (1952) 91f. 7 The figure 1247 does not include what may be a fragment of a gloss preserved in the JOHN J. KEANEY 417 paratively low, is relevant only to the present state of the text: originally it will have been considerably lower. I list the letter-series by ascending number of glosses. The list is followed by three sections which discuss minor dislocations (II), deliberate dislocations and special criteria of alphabetization (III), and major dislocations (IV). LETTER-SERIES NUMBER OF GLOSSES GLOSSES DISLOCATED 'l' 3 0 Z 4 0 ,... .!:!, 6 0 P 9 0 Y 15 0 X 15 1 H 20 3 N 21 1 r 23 1 B 27 2 I 27 1 f[J 33 3 e 34 1 A 34 1 T 36 3 M 46 4 0 51 7 1: 62 9 L1 87 13 K 104 11 II 131 32 E 178 10 A 281 19 1247 122 margin of one MS of n. It is also entirely possible that entire glosses have been lost in the transmission: there are only three glosses in the If'-series, and the Q-series is not repre­ sented. The figure 122 is intended as a maximum. The alphabetic order is affected by some emendations by Dindorf (the latest editor) and others, as well as by corrections which should be made in Dindorf's text. I have omitted from the statistics three glosses which are later interpolations (cf J. J. Keaney, "Moschopoulos and Harpocration," TAPA 100 [1969] 201-07). Because of the special problems involved in the sequences discussed in section IV, the marking of dislocations there is only exempli gratia. 418 ALPHABETIZATION IN HARPOCRA TION'S LEXICON II The MSS of the Lexicon are neither very old nor very good, and it is not surprising to find that defective transmission of the text has affected the alphabetic order of the glosses as well as other elements of the text.8 The clearest evidence is in a sequence like: 1 'Apyck 9 'ApS1]'TT(k 2 >lApyovpa 10 'APE-n] 3 ' APYVPLOO~K1] 11 >I ApO J-LWC 4 ' APYVPOK07TEtOV 12 'AptCTEVC 5* ' ApyLvovcaL om. Ep 13* ' Ap{cTv)J.a om. Ep 6 ' ApYVP07TOVC 14 'AptCT{WV 7* 'Apyo:toc om.Ep 15 ' APKTEVCO:L 8* ' Apwj3ap'aV1]c om. Ep At some early stage in the transmission the dislocated glosses were omitted in the copying, but their omission was not noted until the full page was written. They were then written in the margin, ignored by the epitomator, and reinserted, in the wrong order, into the text of the fuller recension.9 Of nineteen dislocated glosses in the A-series, no less than ten are omitted by Ep. Curiously enough, dislocations which reveal precisely this process are very infrequent elsewhere. The evidence of the A-series shows that glosses were omitted in the course of copying-omissions perhaps facilitated by the very fact of alphabetization-and reinserted into the text. This evidence is bol­ stered by instances in which only one of the two recensions has preserved the alphabetic order, e.g. in the sequences: 1 EtJ-LvAoc 2 EtJ-LwV 1 T1]AEcPaV1]c 2 Tfjvoc In both, Ep has the correct order, n reverses it.l0 If there is not a rep­ etition of the process just described, it means that the scribe of the archetype of n noticed that he had omitted a gloss and immediately repaired the omission. Now, if certain sequences are excluded for 8 In the following lists dislocated glosses are marked with an asterisk. 9 Some of these glosses, e.g . • Apy£voiicQ.L, • ApLofJap'tfVTJc and • AplCTV>J.a, each of which lacks the usual reference to an orator, will have been reinserted in an abbreviated form. 10 Similarly, in the sequence 1 IIpVTav£ia, 2 IIpVTav£lac, II has the correct, Ep the reverse order. JOHN J. KEANEY 419 special reasons,l1 of sixty-three dislocated glosses in all series, forty­ one are out of order by one place, nine others out by two, six out by three, and seven out by more than three. In such cases, it is easy to suppose scribal error: that is, in a sequence like 1 rparfol} 2* rpV1TCX,VLOV 3 rpvlJ..oc it is possible to argue that Harpocration's original order was 1, 3, 2. Indeed in these and other instances in which glosses are out of order by more than one or two places, a variety of the marginalia theory seems possible. In a sequence like 1 Baccxvoc 2 * BanALI<~ 8ux8p0!-,l} 3 BaclA€toc C'TOa 4 BacKa{vEt 5* Bacavlccxc it could be argued that 5, once omitted, was put in the bottom margin and thereafter acquired a new position in the text.12 One is tempted to speak of the probability of scribal error for these minor dislocations, since Harpocration's attempt at complete alpha­ betization was largely successful. On the other hand, it is just these minor slips which might have been committed by the author him­ self, and the evidence of P.Ryl. 532 lends some support to this assump­ tion. It contains portions of two columns, each of which preserves parts of two glosses: 1 Ka'Ta'ToJL~ 1 KE{3pijva 2* Ka'Ta1TA~t 2 KEYXp€OJV The dislocation in the first set is found also in the mediaeval MSS, in which 2 is out of order by three places: 1 Ka'TaC'TaCLC 2 Ka'T(x ~v ayopav atfEv8EI.V 3 Ka'Ta'To!-,~ 4* Ka'Ta1TAl}t 11 I exclude the sequences discussed in section IV, in the first three paragraphs of section III (with n.l5) as well as the entire A-series. Of the nine dislocations in that series not covered by the hypothesis of section II, seven have other textual origins, one is out of order by one place and the last is out by three places. 12 I suspect that Tavpeac, out of order by three places, was similarly omitted and restored in the top margin and in an abbreviated form (it lacks the orator reference). It now stands as the first gloss in the T-series. 420 ALPHABETIZATION IN HARPOCRA TION'S LEXICON Since the papyrus is contemporaneous, or nearly so, with the date of the Lexicon and there was thus not much time for error to have crept in,13 it is safest to assume that Harpocration occasionally erred.a In sum, unless the evidence of one or both recensions clearly indicates scribal error, there is no certain way of telling whether minor disloca­ tions are due to Harpocration or to the transmission of his text. III In at least two instances it is clear that Harpocration himself broke the order for reasons of convenience. The first is 1 B£cIL6c 2 * 8£cILoOlTCXL Here, material from 1 is repeated in 2 and 2 contains a reference to 1 (we 1TpOE{1TOP.EV). The second instance is 1 'Opy£wvac 2* 'Opy'wv Since he had defined the term in 1, the order is necessary, for his only comment in 2 is that Lysias used opyewv instead of OpyEWVWV. opyewv, elsewhere unattested, is probably a lectio falsa which he read in his text of Lysias (cf. LSJ S.V.).15 I have included as dislocation two glosses which probably should not be included. In the sequence 1 "'HCILEV 2 * 'HL77JIL'V'1Jv 3 'H¢>alcna 4* 'Hu.Ov 2 and 4 break the order only if the iota was felt to count for alpha­ betization: since iota-adscript was not pronounced at this time, it probably did not. In the sequence 1 1:vILILopla 2* 1:vyypa¢>£"ic 3 1:VVSLKOL 4 1:vvf]yopoc 5* 1:VyKArrroc €KKATJcla 6 1: vV'Ta~ LC 13 The papyrus preserves a better text than that of the MSS: cf M. Naoumides. "The Papy­ rus of the Lexicon of Harpocration," TAPA 92 (1961) 384-88. U It is conceivable that some dislocated glosses represent marginal additions by Harpo­ cration to his autograph, but nothing is known about the composition of the work. 15 Similarly, the vox nihili 'A'7Tpo7'wv, which Harpocration attempts to emend s.v., was presumably in the text of Dinarchus he was using. JOHN J. KEANEY 421 items Z and 5 would not break the order if the first element of each were written cvv- and normalized later in the tradition: if so, these will have been the forms which Harpocration found in the texts of the orators he was using.I6 There are two instances of dislocation in which the first word of a glossed phrase is repeated: 1 • E7TLJ-LE'AYJTT]C TWV fLVCTYJplwv 2* • E7TLJ-LE'AYJTT]C EfL7TOpLov 1 NEfLEac xcxpaSpcx 2* NEJ-LE&'C It is possible that Harpocration did not feel it strictly necessary to carry alphabetization beyond the first word. Elsewhere, however, alphabetization is carried to two and, in one instance, to three wordsP IV Within the letter-series LI and II there are sequences of glosses with major dislocations, most of which cannot be explained by any of the possibilities suggested in the two preceding sections. These are so puzzling that it is perhaps worthwhile to list them in toto. A 1 LI t&.YPCXfLfLCX 11 LI LaXHPOTOV{CX 2 LlLCXYPCX-rJ 23 IIapeSpoc 16* IIapa:YY€Ala 24 IIape'iaL o4>EtC 17 IIapaKATJcLc 25 IIappTJdac 18 IIapaAoc 26* IIapaKpovE'TaL 19* IIapa{3vcTov 27* IIapacK-rJvLa 20 II apa7T€7T'TWKWC 28* IIapacTJfLoC p-r]TWP 21* IIapaKaTa{3oA-rJ 29* II apaC'TaCLC 22* IIapa{3u>"oLTo 30* II apa4>pvKTwpe tV C 1 II€SLaKa D 1 IIoSocTpa{3TJ 2* II€SapL'Toc 2* IIoSoKaKKTJ 3 II€'ETatpoL 3* IIoALTEla 4 IIEAavoc 4 IIoMfLapxoc 5 II eVTaKocLOfLES'fLVOv 5 IIoMoxoc K'TA. 6 II E7TapTJ(}oc 6 IIofL7Te{ac K'TA. 7* II€{cwv 7 IIopLOc 8* II€vEc'TaL 8* IIoP(JfLoC 9* II €VTTJKoVTapxoc 9* IIoAVyvw'TOC 10* II OAVC'Tpa'Toc 11 IIoc€LS€WV I have been able to discover no satisfactory explanation for these dislocations. They cannot be due to misplacing of folia in a codex18 nor, it seems, are they to be explained by anything in the physical make-up of the roll. No series has lemmata drawn exclusively from one speech or orator: in no series do lemmata of one category (e.g. financial or legal terms) so predominate as to provide an explanation. While it is clear that some of the causes of dislocation discussed in previous sections may apply here (e.g. B 9 is found in five different positions in the MSS), the application would serve only for individual items, and a more general explanation seems called for. It is notable that in all the sequences there are items which are dislocated when viewed as part of the whole series but are alphabetized within sub­ groups (e.g. A 1-6, 7-11, 12-14; B 11-15; C 2-6, 7-9; D4-7,8-11). We have, of course, absolutely no information on Harpocration's method of working, but he must have used some system analogous to our system of file cards.19 It is possible that these series represent an early 18 In all the MSS of the full recension except one, a long sequence of glosses CEKq,V>.A0rP0- pijcru-' EP.7TO'8wV) is displaced, but it is easy to see that this happened because of the displace­ ment of a folium. 18 Daly 85-90 argues that file cards were not used in alphabetized:lists. JOHN J. KEANEY 423 stage of parts of the Lexicon which never were finally organized. It is odd, however, that similar dislocations are not found in the rest of the L1 and n series nor in other series which are longer than these. To conclude. Harpocration (s.v. 'EgoVA:'lC) refers to Caecilius of Calacte, a contemporary of Augustus: the reference is doubtless to Caecilius' rhetorical' Et