FOWLER, ROBERT L., Sappho fr. 31.9 , Greek, Roman and Byzantine Studies, 28:4 
(1987:Winter) p.433 

Sappho fr.31.9 

Robert L. Fowler 

<l>alv~Tal /J-Ot KijVOS LUOS (J£ounv 
)' ,,' )' ., I , 

~/J-/J-~V ooV7IP, OTns ~vavnos TOt 
,~, I \' 9~h. ' to"vavu Kat 7Tl\aO"tOV avv ",wvu-, , 

O"as V7TaKOVft 
, \'" I,'i'\ 

KaL y~l\atO"as t/J-~PO~v, TO /J- 71 /J-av 
~ " '(J " KapvtaV ~v O"T7I ~o"W ~7TTOatO"~v' 

, I " , ,'~ ~ , ", h. ' WS' yap <~s> 0" woo fJPOX~ WS' /J-~ ",oov7l-, ,~, )"" 
0" OVv~v ~T ftKft, 

aAACt tKavt /J-£V yAwO"O"a twyd, A€7TTOV 
~,,' ~ ~, ~ ~ , 
v aVTLKa XpooL 7TVP V7TaV~Vp0/J-aK~V, 
" ~,'~I" "~' O7T7TaT~O"O"t v OVv~V 0P7l/J-/J- , ~7TtfJPO-

~, " /J-ftO"L v aKovaL, 

a O£ /J-' L'opoos KaKX£~TaL, TPO/J-OS' Ot 
~,' \. '~I' 7TaLO-av ayp~L, X l\oopOT~pa v~ 7TOtas 

" 8' ~, '\. ' , ~, 
~/J-/J-t, T~ vaK7IV v Ol\tyoo 7Ttv~V7IS' 

"" I """ [ ",aw0/J- ~/J- aVT at. 
'\.\.' I '\. "t t al\l\a 7Tav TOI\/J-aTOV, ~7Tft KaL 7T~V7ITa 

1-8 vide app. crit. apud ed. Voigt. 9 elAAa K(iV cod. P ps.-Longini 10; 
elAACz KCz,..,. codd. apogrr., recc.; <lAACz KaTCz Pluto Mar. 81D; KaTCz Aneed. 
Par. I 399.27 Cramer. YAwuua {aYE A'TrTOV 0' Long.; yAWUUo. yE 
A'TrTOV Plut.; YAwuuav YEAOTr A need. Par. 10-12 vide Voigt. 13 elOE,..,.' 
l.0pWS KaKbs X'ETaL Epimer. Hom. ap. A need. Ox. I 208.15 Cramer 
(fonte nimirum Herodiano, ut viderunt Lentz, al.); 'KaoE ,..,.' (EK OE ,..,.' 
apogrr.) l.0pWS "'Vxpos K' elKX'ETaL Longin.: textum hic et olim recep­
tum, ab edd. recc. tamen spretum, vindicat Di Benedetto, Hermes 
113(1985) 151-53. 

The purpose of this note is to demonstrate that, in all probability, 
aAA' lJ.Kav /J-Ev yAwuO"a 7T£7TaYf is the correct text of Sappho fr.31.9 
(Lobel-Page=Voigt). At the moment the established vulgate is KCt/J- /J-€V 

... tay~, "my tongue is utterly broken." Various objections had once 
been made against this reading, but it seems that they have been 
overcome. Page objected to the sense, thinking that yAwO"O"a could not 
be used to mean 'power of speech', and that KaTayvv/J-t could not be 

433 



FOWLER, ROBERT L., Sappho fr. 31.9 , Greek, Roman and Byzantine Studies, 28:4 
(1987:Winter) p.433 

434 SAPPHO FR.31.9 

used metaphorically; 1 the first argument was answered by L. E. W ood­
bury, the second by B. B. Ford and E. Christian Kopff.2 The hiatus, to 
which many had objected, was defended by supposing influence of 
Ionic or Aeolic epic,3 or by supposing that Sappho aimed for ono­
matopoeic effect.4 The case for raYE was strengthened by reference to 
Lucretius 3.155, in/ringi linguam, in a passage evidently imitating 
this one of Sappho. One text available in the first century B.C., there­
fore, read raYE; although many of the most serious corruptions in our 
texts occurred at an early date, this evidence must be given due weight 
in the absence of an earlier recoverable text. 

Scholars have not said much about another possible objection 
against the vulgate: the tmesis of lCap. • • • raYE. In his edition of 
Sappho, Lobel reports Cobet's conjecture lCap. •.• 7rE7raYE and remarks, 
"probabiliter, si abesset lCap.."s Since I can see no objection to the 
sense of lCaTa7rf.7raYE, I assume Lobel was objecting to the tmesis, as, 
indeed, M. Treu and G. Privitera assumed.6 The facts with regard to 
tmesis in the Lesbians are these. In Sappho, there are only three 
examples, all conforming to the pattern preverb-postpositive (OE)­
verb (frr.42.2, 48.2, and 168B.3 Voigt, the last being a poem of doubt­
ful ascription; I exclude the adverbial use of EV at fr.2.5, 9). This is a 
very easy type of tmesis, possibly even surviving in colloquial Attic of 
the fifth century, long after all others had become obsolete.7 In AI­
caeus the situation is different: in addition to numerous examples of 
the kind found in Sappho (frr.115.11, 15; 130.18; 338.5f; 346.2; 
362.3; 367.2; see also Inc. Auct. 30.5 Voigt=Alc. fr.255.5 L.-P.), he 

1 D. L. Page, Sappho and Alcaeus (Oxford 1955) 24. 
2 L. E. Woodbury, TAPA 86 (1955) 36; B. B. Ford and E. Christian Kopff, "Sappho 

fr.31.9: A Defense of the Hiatus," Glotta 54 (1976) 52-56. 
3 Homer gives lCaTcl 8' lipp.aTa li.fw, II. 8.403, and E! aVXfva li.fy, 5.161; Hesiod 

provides an exact parallel with E'IT& VOOTa fa-Yf, Op. 534. See R. Hiersche, "Zu Sappho 
2,9 D.," G10tta 44 (1966) 1-5; Ford and Kopff (supra n.2); A. M. Bowie, The Poetic 
Dialect of Sappho and Alcaeus (New York 1981) 85. Ford and Kopff show that it is 
not permissible to speak of lingering influence of the digamma; linguistically, a 
phoneme is either in a dialect or it is not. However, Bowie's formulation is accept­
able: "One might ... say that it is merely a [Aeolic] poetic form preserved after the 
loss of the digamma in the spoken language by the same process that led to such 
hiatus in [Ionic] epic." 

4 Ford and Kopff (supra 0.2); G. Nagy, Comparative Studies in Greek and Indic 
Meter (Cambridge [Mass.] 1974) 45. 

5 E. Lobel, I:a'ITc#>ovS' p.f). • ." (Oxford 1925) xxxii, 17. 
6 M. Treu, Sappho4 (Munich 1968) 178; G. A. Privitera, QUCC 8 (1969) 40 (=La 

rete di Afrodite [Palermo 1974] 89). 
7 J. Wackemagel, Vorlesungen aber Syntax IP (Basel 1928) 172f; but see also G. C. 

Horrocks, PCPS 206 (1980) 1-11. 



FOWLER, ROBERT L., Sappho fr. 31.9 , Greek, Roman and Byzantine Studies, 28:4 
(1987:Winter) p.433 

ROBERT L. FOWLER 435 

affords others of a kind familiar from epic: frr.10.4, E'7I'L yap '7I'apos 
<a>v{aTov llC<aVEL>; 129.3f, EV Of {3WP.OLS •.. f8TJlCav; 208.6, '7I'fP P.fV yap 
" 'I.' 'St" 298 26f ' St" '" I , t I , I 8 '\ \ aVTI\OS LO'"T07f€uaV €X€L; • , €I( u a't'aVTOLS €~a7fLVas €l(VKa V€l\l\aLS; 
376, EIC Of '7I'OT~PLOV 7fWV7]LS; 377, flC p.' EAaO'"as aAyfwv. Another possible 
example is found at fr.336, where Voigt prints '7I'ap.'7I'av Of TVl/JwS flC F' 
"'I. '" I h·l L b I P . t I St' , I", " ~"'" ~A~TO 't'p~vas, W 1 e 0 e - age pnn '7I'ap.'7I'av u ~TV't'Wcr £IC u ~A~TO 

.ppJvas. Since we have no true example of tmesis in Sappho, not even 
in fr.44, tmesis is highly suspect in fr.31.9. 8 

The reading '7I'f.'7I'ay~ (or, more precisely, '7I'f.7f7]y~) was first proposed 
by Joshua Barnes, as reported by C. Blomfield.9 Blomfield himself 
wrote FfFay~. C. G. Cobet, Mnemosyne 2 (1873) 362, again suggested 
'7I'f'7l'ay~, referring to Catullus' translation in poem 51, which gives 
lingua sed tor pet. Both Blomfield and Cobet continued to read lCap. at 
the beginning of the verse. D. L. Page (supra n.1: 25) reports Lobel's 
observation that codex P of Longinus in fact gives us not aAAalCap. as 
the cursus litterarum but aAAalCav, which should probably be articu­
lated aAA' ll.lCav. Noting that this could go well with '7I'f.'7I'ay~ to mean 
"my tongue is fixed in silence," Page states pessimistically that "the 
available MS. evidence does not suffice for a certain restoration," and 
declines further argument. More recently, G. Devereux found that 
'7I'f.'7I'ay€ well suited his psychoanalytical treatment of the poem, and 
gave his support to the conjecture mainly for that reason. 10 

There is, in fact, good evidence that suggests the existence of a text 
reading '7I'f.'7I'ay€ contemporary with or earlier than Lucretius' text with 
fay~. I do not mean Catullus, who, as Page observes, is too free with 
other details of his translation to be relied on for this one. There is, 
however, at least a loose resemblance between his tor pet and our 
'7I'f'7l'ay~; if other testimony suggested the latter, Catullus' testimony 
could be added to theirs. In other words, it does not have independent 
value, but it can have complementary value. This testimony is found 
in Theocritus and Apollonius, in two passages imitating Sappho fr.31. 
In the first, the resemblance has long been noted, but so far as I know, 
only one person has used it to draw inferences about the text of 

8 The situation with respect to tmesis is similar to that with respect to epicisms 
generally, at least as Lobel established it: in Sappho, epicisms are (speaking simply) 
confined to a few poems, whereas in Alcaeus they are scattered evenly throughout his 
oeuvre. The works of A. M. Bowie (supra n.3) and R. Janko, Homer, Hesiod and the 
Hymns (Cambridge 1982), have done much to challenge Lobel's conclusions, but the 
matter is not yet settled. 

9 "Sapphonis Fragmenta," Museum Criticum (Cambridge Classical Researches) I 
(l813, repro 1826) 9 (=T. Gaisford, Poetae Minores Graeci lIP [1823] 295). 

10 G. Devereux, CQ N.S. 20 (1970) 24. 



FOWLER, ROBERT L., Sappho fr. 31.9 , Greek, Roman and Byzantine Studies, 28:4 
(1987:Winter) p.433 

436 SAPPHO FR.31.9 

Sappho's ninth verse, and then only with diffidence. I I In the second, 
the parallel is occasionally noticed in apparatuses, but only one schol­
ar has said more than that, and he was interested in literary, not 
textual, applications.12 To take Theocritus first, in Idyll 2.104-11 he 
describes the moment when Simaetha first sees Delphis entering her 
house: 

, ,~, t, I 
E)'W uE VLV ws EV01'/ua 

" e ' • \ , li> \ , Q I li>' I A. apn vpas V7TEP OVuov ap.ELfJop.EVOV 7TOuL KOV..,..WL-
A. ,/". ' \" e' t'e t' ,~ \. ' ..,..pa~EO P.EV TOV EpW 0 EV LKETO, 7TOTva L.El\aVa-
A" ./,' e ' " ' 'li> \ , 7Taua P.EV E,#,VX 1'/V XLOVOS 7TI\EOV, EK uE P.ETW7TW 

~5::: I '5:::" , , , 
LUpWS P.EV KOXVuEUKEV LUOV vonaLULV EEpuaLS, 

O~O, n <pwvfjuaL ovvcip.av, ovo' ()UUOV lv ii7TVWL 

KVv(EvvTaL <pwvEvvTa <plAav 7TOTL p.aTEpa TEKva, 
, \. \.'" li> Ali> ,,\ , , e " al\l\ E7Tay1'/V ua)'VuL Kal\OV xpoa 7TaVTO EV Lua. 

The passage has much in common with Sappho's: the moment of 
seeing the object of erotic interest is highly dramatized; the list of 
symptoms is extensive and is presented in paratactic style; certain 
items in this list are shared by both authors; and the helplessness of 
the victim is stressed. There is no doubt that Theocritus had Sappho 
in mind when composing this passage. When we inquire just what 
presence Sappho's poem has in Theocritus' final composition, the 
answer is interesting. The underlined portions indicate how key ideas 
have been borrowed from Sappho (lv61'/ua has been substituted for 
roov, KOXVOEUKEV for KaKXEETaL, and ovvap.av for ErKEL); the non-under­
lined portions indicate how they have been elaborated. I do not 
suppose that Theocritus actually composed in so mechanical a man­
ner, but this is the end result; nor is it less interesting for being 
susceptible of such articulation. In fact, understood in this way, the 
passage presents a delightful counterpoint as Sappho's simpler ex­
pressions are one by one answered by Theocritus' inventive vari­
ations. Particularly exquisite is the addition of the doll, which won­
derfully retains a suggestion ofSappho's muteness. 13 

11 Devereux (supra n.lO). He notes in passing Theocritus' haY71v, but is uncertain 
about its implications for Sappho's text. He offers no opinion on the first words of the 
line. 

12 Privitera (supra n.6) 72 (=La rete 123). 
13 Line 105 contains an allusion to Aesch. Ag. 408f, which some have thought 

imitates Sappho fr.16.12f (W. M. Calder III, "An Echo of Sappho Fragment 16 L-P at 
Aeschylus, Agamemnon 403-419?" Apophoreta phil%gica Emmanueli Fernandez­
Galiano a sodalibus ob/ata, edd. L. Gil, R. M. Aguilar, I [Madrid 1984] 215-18; B. 
Marzullo, Museum Criticum 13-14 [1978179] 108). Line 107 might contain an 
allusion to a text containing the word 'itvxpos in line 13. The detail is suggestive, but 
perhaps cannot be pressed with respect to the true reading in Sappho; it is true, 



FOWLER, ROBERT L., Sappho fr. 31.9 , Greek, Roman and Byzantine Studies, 28:4 
(1987:Winter) p.433 

ROBERT L. FOWLER 437 

If this analysis is correct, it has implications for Sappho's text. Note 
in particular the sequence OVOf Tt cpwvijCTat ovvap.av .•. aAA' €7TaY1Jv, 

d · t" '" ' " !>o \ ",,, , \ \ \ , Th correspon lng 0 ws P.E 't'WV1JCT OVuEV ET EtKEt, al\l\a ••• 7TE7TaYE. e 
intransitive present perfect has been changed to an intransitive sec­
ond aorist, since Theocritus' narrative is set in the past; otherwise it is 
a quotation. 

It is possible, of course, that Theocritus' variation extends to in­
clude 7Tf7TaYE. One scholar has suggested that the detail is an elegant 
example of oppositio in imitando, opposed, that is, to TPOP.OS; 14 anoth­
er, noting the sequence OVO~(V) •.• aAAa as I have done, thinks that 
7Tf7TaYE may have been suggested to Theocritus by the sound of faYE. 15 

Neither one considers an alternative reading in Sappho. To my judg­
ment, what is argued above strongly suggests that Sappho's text had 
7Tf7TaYE; but I admit that these other suggestions are still possible. Let 
us then tum to Apollonius 3.962-65, describing the first meeting of 
Jason and Medea: 

, !>o'" , !>o' (J' , " !>o' " 
EK u apa Ot Kpaut1J CTT1J EWV 7TECTEV, op.p.aTa u aVTWS 
" \ (J \ !>o \ '!>O ~\ " (J 1JX I\VCTaV, EPP.OV uE 7Tap1Jtuas EtI\EV EpEV os' 

, ~,,' ,'),), '(J' ~ 
yovvaTa u OVT 07TtCTW OVTE 7Tp07TapOt EV aEtpat 
"(J '\\'" (J' '!>o ECT EVEV, al\l\ V7TEVEP E 7TaY1J 7TOuas. 

The relationship with Sappho is less close than it was in the case of 
Theocritus, but there is a relationship nonetheless: we have the same 
general situation, and stress on the moment of sight; we have the same 
paratactic list of symptoms; we have the same stress on love's over­
whelming impact on the victim's faculties. A skeptic might contend 
that by this time the pathology of love was conventional, and that any 
poet setting out to describe such situations would compose passages 
like these. But this account would not do justice to the extent of the 

though, that Di Benedetto's explanation for the occurrence of "'vXpos in some testi­
monia is the weakest part of his case (he assumes it is a gloss on KaKOS, which was 
itself extracted from KaKxEfTaL). Yet other texts must assume a process of corruption 
more complicated than his, and he is right to stress the value of Herodian's 
testimony. Unlikely though "'vxpos may seem as a gloss for KaKOS, in this context KaKOS 
is obviously impossible, while 'cold' is a very probable adjective; perhaps instead of 
'gloss' we should speak of 'conjecture'. If Theocritus' text of Sappho read something 
like a aE p.' rapws "'vxpos iKaKiXEfTaL, unmetrical though that is, we should say that he 
expanded the line into two statements about cold and sweat respectively, with extra 
details for each; in this case, ?faUa p.tv f."'~X81jv ought also to be underlined above. My 
argument regarding 7TE7TaYf remains unaffected in any case. 

14 R. Pretagostini, "Teocrito e Saffo: forme allusive e contenuti nuovi," QUCC 24 
(1977) 107-18. 

15 V. Di Benedetto (supra 433) 154 n.27. 



FOWLER, ROBERT L., Sappho fr. 31.9 , Greek, Roman and Byzantine Studies, 28:4 
(1987:Winter) p.433 

438 SAPPHO FR.31.9 

similarity, nor to the intimate knowledge of classical texts possessed 
by Apollonius. 

In this light it is highly suggestive that Apollonius uses the verb 
'1Tay,., in line 964, even if the application is different. He also has the 
S h· " ( ) " 8 (' lit \ "'" , lit \ lit ' ) app IC sequence OVT € ••• HI' €V€V -OVu€V €T €&1C€&-OVu€ uvvalJ-av , 
aAA' ... '1Tay,.,. I only say that these items are suggestive, particularly 
since Homeric influence is strong in lines 964f; they may be inspired 
entirely by that author. 16 But let us now recall Catullus. Whatever one 
may think about the precise relations of Homer, Sappho, Theocritus, 
Apollonius, and Catullus, the last three do allude with varying degrees 
of closeness to Sappho, and Catullus and Theocritus had their eyes 
directly on her text; Catullus has tor pet, Theocritus has £'1Tay,.,v. Apol­
lonius has '1Tay,." which, depending on how one understands the gene­
sis of these lines, mayor may not be relevant. Theocritus and Catullus 
together are really enough to tip the balance in favour of '1TE'1Tay€, but 
Apollonius' lines are most curious in the light of their agreement. 17 

Given that we have also found a new reason to object to lay€ (tmesis), 
the chances are very good that '1TE'1Tay€ is right. It seems, in fact, to 
have been the reading of the text in the Alexandrian library of the 
third century B.C. and is therefore preferable to Lucretius' version. 

We still have the beginning of the line to consider; lCaTa'1TE'1TaYE is 
unobjectionable in sense, as I have said, but the tmesis continues to 
offend. aAA' l1."av is, however, practically given to us by the archetype 
of pseudo-Longinus, and gives good sense}8 Accounting for the cor-

16 Parallels in M. Campbell, Echoes and Imitation of Early Epic in Apol/onius Rho­
dius (=Mnemosyne Suppl. 72 [Leiden 1981)) 57; note particularly II. 10.94f for line 
962, 22.452f for 964f. 

11 I am bound to mention the curious occurrence of the (corrupt) expression 
cTT7J8'wv taY7J /dap at line 954 of Apollonius, very shortly before the passage here 
under discussion. If fc1Y7J is really there, its presence seems to be purely coincidental, 
unless we suppose that the Alexandrian text had the variant fayE in it (perhaps 
written above the line); ApoUonius in that case would be sitting on a philological 
fence by alluding to both readings. I have assumed throughout that the variant is 
attested by Lucretius, but it should be noted that the context in his poem is different, 
and some authorities have doubted Sappho's influence: see E. J. Kenney's note ad 
154-56 in his commentary (Cambridge 1971). For the typically Alexandrian allu­
siveness I assume in ApoUonius (and for the way in which obscure textual variants 
can be incorporated into poems) see e.g. G. Giangrande, "'Arte Allusiva' and Alex­
andrian Epic Poetry," CQ N.S. 17 (1967) 85-97 (=Scripta Minora Alexandrina I 
[Amsterdam 1980] 11-23), and "HeUenistic Poetry and Homer," AC 39 (1970) 46-77 
(=33-64); E. Livrea, "Una tecnica allusiva apolloniana alla luce deU'esegesi omerica 
alessandrina," StIt 44 (1972) 231-43. lowe these references to C. G. Brown, who also 
points out that H. Frankel detected a further allusion to Sappho fr.47 at 3.972, 
shortly after our passage: Noten zu den Argonautika des Apol/onios (Munich 1968) 
409. 

18 Privitera (supra n.6) 41 (=La rete 89) calls l1ICav a "zeppa" or filler, but epic finds 



FOWLER, ROBERT L., Sappho fr. 31.9 , Greek, Roman and Byzantine Studies, 28:4 
(1987:Winter) p.433 

ROBERT L. FOWLER 439 

ruption does not seem an urgent requirement; the fragment itself, 
which is generally in a wretched state, offers other corruptions that 
have arisen from omission of letters and syllables. aAA' l1Kav JJ.EV 
-yAwuua 7r'7ra-YE is the oldest recoverable text; it is unobjectionable in 
sense, syntax, dialect, and metre; it should be printed. 

We may take it, then, that the text offr.31.9, having been broken, is 
now fixed. 19 

UNIVERSITY OF WATERLOO 

November, 1987 

a."~V £}'£VOVTO cnw7Tfi a perfectly acceptable formula (e.g. II. 3.95, 7.92); indeed, 
Sappho may have had just this expression in mind, since it is used "plerumque de iis 
qui affectu aliquo perturbati obmutuerunt" (H. Ebeling, Lexicon Homericum s. v. 
a.,,~v): a good description of Sappho herself. C. del Grande, Ant%gia della Urica 
greca 2 (Naples 1959) 116, cites Hor. Carm. 4.1.35f (parum decoro ... cadit lingua 
silentio) as a parallel for l1."av, although the stress laid on silentio by the adjectival 
phrase parum decoro renders the parallel imperfect (silentio strikes the modern ear as 
less of a "zeppa"'). 

19 Professor C. G. Brown provided valuable criticism of the draft of this paper, for 
which I thank him.