In HN 9.126, Pliny describes the extraction of purple from sea snails. Modern editions print the following text of the passage:Footnote ¹

The words cum testa uiuas, however, are a conjecture by Jean Hardouin based on the reading cum terra uitis preserved by most manuscripts.Footnote ² It is clear that they roughly correspond to the sense required,Footnote ³ since the passage is dependent on Arist. Hist. an. 547a21–7:

Compare, in particular, μϵτὰ τῶν ὀστράκων in Aristotle’s text. But, while Hardouin’s conjecture is generally appropriate and might seem to be additionally supported by the reading of F² (unknown to Hardouin), cum terra uitas,Footnote ⁴ it is not really easy to imagine why testa should have been corrupted to terra or uiuas to uitas/uitis.

At the same time, Michael Reeve has shown in his recent monographFootnote ⁵ that in Books 1–13 of the Natural History a manuscript family alternative to the one better known to the editors is available even where its readings were earlier thought to be lost, as in the passage in question. The best continuous representative of this family is MS P (Prague, Národní knihovna České republiky, XIV A 12, fifteenth century), not used by modern editors before Reeve, and it is convenient to check its readings with the second-hand corrections in MS L (Florence, BML, Plut. 82.1–2, twelfth century). Now both P and L² give a different variant instead of cum terra uitis, viz. contritis conchis. The readings of the family were actually known to some early editors such as S. Gelenius,Footnote ⁶ and the editio princeps actually printed the text contritis conchis.Footnote ⁷ However, other fifteenth- and sixteenth-century editors who were aware of this reading seem to have tacitly rejected it as an obvious conjectural interpolation.Footnote ⁸ Still it might be sensible to reconsider this assessment in the light of Reeve’s evaluation of the family.

What is interesting about contritis is that it can be shown to be not just a contextual replacement for the cum terra uitis; the discrepancy clearly results from a mechanical corruption: cum and con- sometimes share the same abbreviation in early medieval scriptsFootnote ⁹ and are thus easily confused; terra is actually abbreviated as tra with a tilda in our place in R, one of the two oldest independent manuscripts of the more widespread family, according to Reeve, and the abbreviation tr with a tilda is also attested in the early ninth century;Footnote ¹⁰ and the letters -itis at the end coincide. Even though I do not know exactly how to explain the appearance of u,Footnote ¹¹ it seems anyway clear that the two variants reflect independently the same text of the archetype and so must be considered on equal terms. Once we compare them thus, contritis (conchis) is certainly preferable: in fact, it gives the very sense of μϵτὰ τῶν ὀστράκων in a way which would even as a conjecture be palaeographically neater than cum testa uiuas. Note also that uiuas is actually superfluous, since it has already been said that all the purples must be used while alive. The change in number from singular to plural is also understandable, since only large purples receive individual treatment while smaller ones have to be crushed collectively (cf. the corresponding use of, respectively, τὸ ὄστρακον and τῶν ὀστράκων in Aristotle). The only reservation that has to be made is that it is only contritis that corresponds to cum terra uitis palaeographically, not conchis; however, the word conchis is not actually necessitated by the context. After another ablative absolute, detracta concha, an elliptical use of contritis for contritis conchis will be very easy to understand.Footnote ¹² conchis can most conveniently be interpreted as a gloss to the elliptical expression that entered the text. Consequently, I suggest that we should read et maioribus quidem purpuris detracta concha auferunt, minores contritis frangunt, ita demum eum expuentes.

One further complication remains to be commented upon. MS F², which, as I have said, reads cum terra uitas, according to Reeve, belongs to the same family as MSS PL² and is a better witness to its text than them, at least when this corrector reflects the readings of its main source. If this is the case in our passage, the reading of PL² must be a later conjectural innovation in the family, since the reading of F² is closer to the reading of the more widespread family. However, if cum terra uitas is the reading of the archetype, the reading of PL² must depend not on it but on the reading of the more widespread family, and this would imply a line of contamination which seems not to be confirmed otherwise; anyway, some readings of F² certainly have to be interpreted as purely conjectural attempts to correct the reading of the ‘vulgate’ without recurring to the readings of the alternative family: cf. for example HN 9.166, where the true reading is clearly acus siue (known from MS R² and some late manuscripts), while P read actis siue; as for the ‘vulgate’ manuscripts, R before correction read something like a cuatue (the letters between the two u are now erased) and the descendants of D (not extant there) read a coitu e, except F, which has a coetu e before correction and a coitu e after correction by the second hand.Footnote ¹³ I suggest that it is most sensible to see in the reading cum terra uitas a similar conjecture, although not a very felicitous one.