From Parmenides four passages are quoted; strangely enough three of them are passages that had been quoted by Plato. Lines 52-53 in our texts of Aristotle repeat the same error that appears in our texts of Plato; ll. 103-105 are not so near to what seems to be the original (judged by the quotation in Simplicius) as is the Platonic version. Unless our MSS. are greatly at fault, two of the four passages were very carelessly reproduced, and we have reason to believe that they were drawn from Plato. The fourth passage, given by Aristotle and Theophrastos, has the appearance of careful quotation, though one verb has an unmetrical form in our Aristotle (where Theophrastos gives a correct form). Aristotle does not quote directly from either Zeno or Melissos. Coming now to Empedokles, we find two extended passages which can only be regarded as genuine quotations, namely ll. 287-311 and 316-325. On the other hand several phrases (ll. 208, 326, 443) give only a general idea of the language of Empedokles. Most of the quotations consist of from one to four lines preserving their metrical form, so that they deserve the name of quotations; but their accuracy is doubtful in matters of detail. This is most clearly seen by an examination of the ten cases where the same passage is quoted twice by Aristotle, namely: lines 36-39, 104-107, 146-148, 167, 208, 244, 270-271, 330-332, 333-335. In only three of these instances (38-39, 270-271, 333-335) is the quotation identical; in the other cases there is some slight difference in the text, although commonly both versions scan correctly. An examination of the lines quoted only once in Aristotle shows very frequent deviation from the same lines as quoted by others. In two instances a line is omitted from the context (37 and 99); a case is changed, a connecting particle changed or omitted entirely, a common word is substituted for a rarer one (236-237) or an Aristotelian word for the word required by the full context (e.g. _Meta._ 1015 a 1), or finally only the substance of the line is given (e.g. lines 91, 92). These variations are so numerous as to justify the conclusion that the text furnished by Simplicius or by Sextus Empiricus deserves quite as much weight as that furnished by Aristotle, since the latter cares only for the thought and not at all for the exact language in which the thought had been clothed.