The tenth and final chapter of the first book of Aristotle's Metaphysics is short, comprising sixteen Bekker-lines, from 993a11 to 993a27.Footnote ¹ Yet, it is not short enough to be free of textual puzzles. In the first six lines (993a11–17), the divergences between the two branches of the direct manuscript tradition (α and β)Footnote ² can be resolved relatively easily or are of minor significance.Footnote ³ There are two textual phenomena, however, that are trickier. These two instances do not come in the form of a disagreement between α and β. Indeed, all our manuscripts read the same flawed text. The section that contains these instances—the first six lines of A 10—is the following:

The first textual problem concerns the words καὶ τὸ πρῶτον in lines 993a16–17. As RossFootnote ⁶ and Jaeger,Footnote ⁷ following Diels,Footnote ⁸ have indicated, these words should be deleted as a gloss. The evidence in Alexander of Aphrodisias’ commentary supports this deletion.Footnote ⁹ The second textual problem, on which I will focus, shares with the first problem the feature that Alexander's text provides a better reading than our manuscript tradition (α and β). This second problem, however, has not even been identified as a problem. It is the purpose of this note to diagnose the problem and then to provide a solution for it.

This problem concerns the sentence in lines 993a13–15, which contains three oddities: one in syntax, one in diction and one in content. First, the syntax. In the first clause of the sentence (ἀλλ’ ἀμυδρῶς ταύτας), neither the verb nor the subject is stated. Both, verb and subject, need to be supplied. They can be supplied easily, because the object (ταύτας ‘these’, that is, the causes) makes it clear that an active verb such as ‘speak’ or ‘seek’ (cf. ζητεῖν ἐοίκασι in a12) with the subject ‘they’, namely Aristotle's predecessors who featured in the previous sentence, is needed. However, in the second clause of the sentence, connected to the first with a καί, we suddenly have a new construction consisting of a verb in the passive voice (namely εἴρηνται ‘have been said’) and a new, expressed subject (namely πᾶσαι ‘all, sc. the causes’), which was the object in the first clause. This sudden and seemingly unjustified switch in construction is odd. This is so not only because the second clause does not provide the verb and the subject that are needed in the first clause but also and especially because the second clause continues with a completely different syntactical construction. This switch is quite impractical. The first clause's object (‘the causes’) turns into the second clause's new subject and the verb turns from an implied but unexpressed active form into an explicitly stated passive form. Meanwhile, the implied but unexpressed subject of the first clause (‘the predecessors’) has completely disappeared in the second clause. There seems no good reason for this switch.

Second, the diction. The phrase πρότερον εἴρηνται meaning ‘they [that is, all causes] have been (generally) said before’ is odd, first, because the phrase ‘said before’ (τῶν πρότερον εἰρημένων) has just been used (line 13) for the quite different purpose of referring back to something Aristotle said earlier and, second, because it is typically used for such back references to Aristotle's own words.Footnote ¹⁰ Here, however, it appears to express in general terms the fact that ‘all’ of Aristotle's four causes have been said or named by some of his predecessors some time before. It is difficult to find parallels for this use of the verb in Aristotle.Footnote ¹¹ Specifically, according to a TLG search, the form εἴρηνται (third-person plural perfect passive, ‘they have been said’, implying a non-neuter subject) occurs only seven other times in the entire corpus. By contrast, the third-person singular form εἴρηται (‘it has been said’), to which I come back below, appears 1,006 times. In six of the seven occurrences of εἴρηνται, Aristotle is the agent of the verb, that is, the place where things ‘have been said’ is Aristotle's own writings;Footnote ¹² in one of the seven occurrences, the subject of ‘have been said’ are proverbs and the agent of the verb is any Greek speaker.Footnote ¹³ Hence, the use of the form εἴρηνται with an agent that is explicitly not Aristotle (but his predecessors instead) is without parallel in the Aristotelian corpus.

Third, the content. The two clauses that make up our sentence seem unbalanced in what they convey. The switch from an implied active to an expressly stated passive construction that happens between the first clause and the second clause results in odd content. For, the syntactical switch causes a strange shift in the sentence's focus. In the first clause, Aristotle focusses on his predecessors and that they speak unclearly about the causes (he delineated in the Physics); in the second clause of the sentence, he states how the causes generally have and have not been named or said before. Whereas in the first clause the focus is on the predecessors, this focus is strangely blurred in the second clause.

I suggest that the reading in our text (α and β) is faulty. Alexander of Aphrodisias’ commentary gives us access to an older version of the Metaphysics,Footnote ¹⁴ and in the given case this older version has a different and, so I want to show, better text. Surprisingly, no Metaphysics editorFootnote ¹⁵ or scholar has recognized that Alexander's Metaphysics text differs from ours in line 993a14. This oversight may at least in part be due to the fact that Alexander quotes lines 993a11–16 not in the section where he comments on A 10 but at an earlier point, specifically in his commentary on A 7.Footnote ¹⁶ According to Alexander's testimony,Footnote ¹⁷ his Metaphysics exemplar read in lines 13–15:

Note the following differences to our text: Alexander's text reads the accusative ἁπάσας instead of the nominative πᾶσαι, and then the phrase ὡς καὶ πρότερον εἴρηται (‘as has also been said earlier’) instead of πρότερον εἴρηνται (‘they have been said before’). The accusative ἁπάσας fits well with ταύτας in the first clause of the sentence; the construction of the first clause continues seamlessly in the second clause. According to this version, Aristotle says that the predecessors seek these causes ‘dimly’ or ‘unclearly’ and that they ‘in one way’ seek all of them. Here, in contrast to the text of our manuscripts, Aristotle inserts a back reference to his own previous words (‘as has also been said earlier’). As I will show presently, this reference is to a specific statement in A 7. Then Aristotle closes, just as in the version of our text, by saying that ‘in another way’ the predecessors seek those causes ‘not at all’.

Why is Alexander's text, compared with the text in our manuscripts, more likely the original text? An immediate answer is because Alexander's text is free of all three oddities that we find in the text of our manuscripts. First, the syntactical oddity of a switch from an implied active verb to an explicit passive construction is healed. There is no switch in Alexander's text, but the verb, implied from the previous sentence, as well as the corresponding subject and object are consistent throughout. Second, and even more importantly, Aristotle uses the formula (ὡς καὶ) πρότερον εἴρηται regularly in other works to refer to an earlier statement,Footnote ¹⁸ while the combination of πρότερον and εἴρηνται (read in the text of our manuscripts) is without parallel. Since we are here at the beginning of the last chapter of Book A, back-references to earlier parts of the book are in order. And we already encountered one in the first sentence of the chapter. Third, there is no odd change in focus of what Aristotle is saying. The entire sentence is consistently about how the predecessors searched for the four causes: in a sense they got them all, yet in another they missed them entirely.

Could, however, the smoothness of Alexander's text be the result of a correction made to the text, either one that aimed at smoothness or one that stemmed from the expectation of a back reference? This possibility cannot be completely ruled out, but it can be shown to be very unlikely for two reasons.

First, the genesis of the error that I claim occurred in the text of our manuscripts can be fairly easily reconstructed. From the original sequence of letters ΠΑΣΑΣΩΣΚΑΙΠΡΟΤΕΡΟΝ resulted ΠΑΣΑΙΠΡΟΤΕΡΟΝ by the following simple mistake: a scribe wrote ΠΑΣ… and instead of continuing the word with the ending …ΑΣ (accusative plural) he jumped a few letters ahead (leaving out ΑΣΩΣΚ) and completed it with the ending …ΑΙ (nominative plural). The letters ΑΙ were originally the last two letters of the word καὶ (ἁπάσ-ας, ὡς κ-αὶ πρότερον —>> πᾶσ-αι πρότερον). To make the resulting sentence intelligible, the verb form was then adjusted from the singular εἴρηται to the plural εἴρηνται.

Second, Alexander's reading makes better sense in the bigger picture, as it opens up new vistas about Aristotle's verdict of his predecessors, the status and function of chapter 10, which had been questioned by Jaeger, and hence even the unity of Book A.

On the basis of the new reading, we can see more clearly what Aristotle is doing at the beginning of chapter 10. First, in lines 993a11–13 (see text above), he starts off by summarizing the results of his discussion in the previous chapters: his predecessors searched for the same causes that he presented in the Physics and for no others besides these. This result strengthens the suspicion he already expressed in chapter 7 (988a20–22), where he gave an interim summary of his discussion in chapters 3–6. Thus, when Aristotle says it ‘is clear from what has been said before’ (993a13), he is referring to what he said at the beginning of chapter 7.Footnote ¹⁹ The relevant lines are:

On the basis of our new reading in line 993a14, we can now see that Aristotle makes another reference to the passage in chapter 7. While the first reference concerns the question whether the predecessors sought a cause outside those stated in the Physics, the new, second reference concerns the question of how and to what extent they sought those causes. Crucially and in contrast to the first reference, the second reference does not simply repeat a point made in chapter 7, but corrects or revises the view expressed in chapter 7.

I will briefly spell this out. According to the new reading that I reconstructed from Alexander's commentary, Aristotle says ‘But [sc. they seek] these in an unclear way (ἀμυδρῶς) and in one way all of them, as has also been said earlier’. This is a reference specifically to chapter 7, lines 988a22–3.Footnote ²⁰ The correspondence is clear owing to the adverb ἀμυδρῶς, which is used in both passages. In chapter 7 (988a22–3), Aristotle states that his predecessors spoke about the causes ‘in an unclear way’ (ἀμυδρῶς) and that they ‘somehow’ or ‘vaguely’ (πως) touched upon the causes. In chapter 10, Aristotle similarly states that his predecessors searched for the causes ‘in an unclear way’ (ἀμυδρῶς) and that in some sense they got them all, but then continues by saying that ‘in another way not at all’ (τρόπον δέ τινα οὐδαμῶς). In between the repetition of the view of chapter 7 and the correction of that very view Aristotle added a back-reference to make sure that the reader recognized the correction as a correction and adjustment of his previous statement. This correction of his own previous view is justified after Aristotle took in chapters 8–9 a second look at certain predecessors, but with a focus slightly different from the one he had in chapters 3–6.Footnote ²¹ The new back-reference in chapter 10 (993a14) shows that Aristotle explicitly adjusts the verdict he reached in chapter 7. By means of this back-reference, Aristotle highlights that the somewhat grim conclusion he offers in A 10 is a correction of the milder view he expressed earlier in chapter 7. After his renewed investigation in chapters 8–9, he must now conclude that in a sense they really did not get it at all.

That Aristotle explicitly refers back to chapter 7 and explicitly revises his earlier interim result makes it possible to rule out Werner Jaeger's influential but misleading thesis about the origin and status of chapter 10. Jaeger argued that Aristotle wrote A 10 as a later version of A 7 with the intention to replace the latter by the former. A later editor then attached what we know as chapter 10 in lack of a better place to the end of the book.Footnote ²² Now that we have at the beginning of A 10 a new back-reference to A 7 that explicitly revises the view expressed in A 7, we have acquired a strong piece of evidence that Book A is not a jumbled patchwork, as scholars have suggested, but that Aristotle composed the book as a series of consecutive steps in a didactic process of fully understanding his predecessors’ contribution to wisdom (sophia). A 7 and A 10 are not randomly placed duplicates but interconnected parts of a finely crafted whole. Aristotle composed A 7 to give an interim report of the results of his discussion in chapters 3–6, and then A 10 to give a full and revised report of the results of chapters 3–6 and 8–9.Footnote ²³ The new back-reference that comes to light through the testimony in Alexander's commentary helps us to appreciate the purpose of chapter 10 within the architecture of Book A.