by A E Housman
This I cite as a specimen of the things which people may say if they do not think about the meaning of what they are saying, and especially as an example of the danger of dealing in generalisations. The best way to treat such pretentious inanities is to transfer them from the sphere of textual criticism, where the difference between truth and falsehood or between sense and nonsense is little regarded and seldom even perceived, into some sphere where men are obliged to use concrete and sensuous terms, which force them, however reluctantly, to think.
I ask this scholar, this critic who knows his business, and who says that the more sincere of two MSS. is and must be the better — I ask him to tell me which weighs most, a tall man or a fat man. He cannot answer; nobody can; everybody sees in a moment that the question is absurd. Tall and fat are adjectives which transport even a textual critic from the world of humbug into the world of reality, a world inhabited by comparatively thoughtful people, such as butchers and grocers, who depend on their brains for their bread. There he begins to understand that to such general questions any answer must be false; that judgment can only be pronounced on individual specimens; that everything depends on the degree of tallness and the degree of fatness. It may well be that an inch of girth adds more weight than an inch of height, or vice versa; but that altitude is incomparably more ponderous than obesity, or obesity than altitude, and that an inch of one depresses the scale more than a yard of the other, has never been maintained. The way to find out whether this tall man weighs more or less than that fat man is to weigh them; and the way to find out whether this corrupt MS. is better or worse than that interpolated MS. is to collect and compare their readings; not to ride easily off on the false and ridiculous generalisation that the more sincere MS. is and must be the better.
When you call a MS. sincere you instantly engage on its behalf the moral sympathy of the thoughtless: moral sympathy is a line in which they are very strong. I do not desire to exclude morality from textual criticism; I wish indeed that some moral qualities were commoner in textual criticism than they are; but let us not indulge our moral emotions out of season. It may be that a scribe who interpolates, who makes changes deliberately, is guilty of wickedness, while a scribe who makes changes accidentally, because he is sleepy or illiterate or drunk, is guilty of none; but that is a question which will be determined by a competent authority at the Day of Judgment, and is no concern of ours. Our concern is not with the eternal destiny of the scribe, but with the temporal utility of the MS.; and a MS. is useful or the reverse in proportion to the amount of truth which it discloses or conceals, no matter what may be the causes of the disclosure or concealment. It is a mistake to suppose that deliberate change is always or necessarily more destructive of truth than accidental change; and even if it were, the main question, as I have said already, is one of degree. A MS. in which 1 per cent. of the words have been viciously and intentionally altered and 99 per cent. are right is not so bad as a MS. in which only 1 per cent. are right and 99 per cent. have been altered virtuously and unintentionally; and if you go to a critic with any such vague inquiry as the question whether the “more sincere” or the “more correct” of two MSS. is the better, he will reply, “If I am to answer that question, you must show me the two MSS. first; for aught that I know at present, from the terms of your query, either may be better than the other, or both may be equal.” But that is what the incompetent intruders into criticism can never admit. They must have a better MS., whether it exists or no; because they could never get along without one. If Providence permitted two MSS. to be equal, the editor would have to choose between their readings by considerations of intrinsic merit, and in order to do that he would need to acquire intelligence and impartiality and willingness to take pains, and all sorts of things which he neither has nor wishes for; and he feels sure that God, who tempers the wind to the shorn lamb, can never have meant to lay upon his shoulders such a burden as this.
This is thoughtlessness in the sphere of recension: come now to the sphere of emendation. There is one foolish sort of conjecture which seems to be commoner in the British Isles than anywhere else, though it is also practised abroad, and of late years especially at Munich. The practice is, if you have persuaded yourself that a text is corrupt, to alter a letter or two and see what happens. If what happens is anything which the warmest good-will can mistake for sense and grammar, you call it an emendation; and you call this silly game the palaeographical method.
The palaeographical method has always been the delight of tiros and the scorn of critics. Haupt, for example, used to warn his pupils against mistaking this sort of thing for emendation. “The prime requisite of a good emendation,” said he, “is that it should start from the thought; it is only afterwards that other considerations, such as those of metre or possibilities, such as the interchange of letters, are taken into account.” And again: “If the sense requires it, I am prepared to write Constantinopolitanus where the MSS. have the monosyllabic interjection o.” And again: “From the requirement that one should always begin with the thought, there results, as is self-evident, the negative aspect of the case, that one should not, at the outset, consider what exchange of letters may possibly have brought about the corruption of the passage one is dealing with.” And further, in his oration on Lachmann as a critic: “Some people, if they see that anything in an ancient text wants correcting, immediately betake themselves to the art of palaeography, investigate the shapes of letters and the forms of abbreviation, and try one dodge after another, as if it were a game, until they hit upon something which they think they can substitute for the corruption; as if forsooth truth were generally discovered by shots of that sort, or as if emendation could take its rise from anything but a careful consideration of the thought.”
But even when palaeography is kept in her proper place, as handmaid, and not allowed to give herself the airs of mistress, she is apt to be overworked. There is a preference for conjectures which call in the aid of palaeography, and which assume, as the cause of error, the accidental interchange of similar letters or similar words, although other causes of error are known to exist. One is presented, for instance, with the following maxim: “Interpolation is, speaking generally, comparatively an uncommon source of alteration, and we should therefore be loth to assume it in a given case.”
Every case is a given case; so what this maxim really means is that we should always be loth to assume interpolation as a source of alteration. But it is certain, and admitted by this writer when he uses the phrase “comparatively uncommon,” that interpolation does occur; so he is telling us that we should be loth to assume interpolation even when that assumption is true. And the reason why we are to behave in this ridiculous manner is that interpolation is, speaking generally, comparatively an uncommon source of alteration.
Now to detect a non sequitur, unless it leads to an unwelcome conclusion, is as much beyond the power of the average reader as it is beyond the power of the average writer to attach ideas to his own words when those words are terms of textual criticism. I will therefore substitute other terms, terms to which ideas must be attached; and I invite consideration of this maxim and this ratiocination:
“A bullet-wound is, speaking generally, comparatively an uncommon cause of death, and we should therefore be loth to assume it in a given case.”
Should we? Should we be loth to assume a bullet-wound as the cause of death if the given case were death on a battlefield? and should we be loth to do so for the reason alleged, that a bullet-wound is, speaking generally, comparatively an uncommon cause of death? Ought we to assume instead the commonest cause of death, and assign death on a battlefield to tuberculosis? What would be thought of a counsellor who enjoined that method of procedure? Well, it would probably be thought that he was a textual critic strayed from home.
Why is interpolation comparatively uncommon? For the same reason that bullet-wounds are: because the opportunity for it is comparatively uncommon. Interpolation is provoked by real or supposed dif
ficulties, and is not frequently volunteered where all is plain sailing; whereas accidental alteration may happen anywhere. Every letter of every word lies exposed to it, and that is the sole reason why accidental alteration is more common. In a given case where either assumption is possible, the assumption of interpolation is equally probable, nay more probable; because action with a motive is more probable than action without a motive. The truth therefore is that in such a case we should be loth to assume accident and should rather assume interpolation; and the circumstance that such cases are comparatively uncommon is no reason for behaving irrationally when they occur.
There is one special province of textual criticism, a large and important province, which is concerned with the establishment of rules of grammar and of metre. Those rules are in part traditional, and given us by the ancient grammarians; but in part they are formed by our own induction from what we find in the MSS. of Greek and Latin authors; and even the traditional rules must of course be tested by comparison with the witness of the MSS. But every rule, whether traditional or framed from induction, is sometimes broken by the MSS.; it may be by few, it may be by many; it may be seldom, it may be often; and critics may then say that the MSS. are wrong and may correct them in accordance with the rule. The state of affairs is apparently, nay evidently, paradoxical. The MSS. are the material upon which we base our rule, and then, when we have got our rule, we turn round upon the MSS. and say that the rule, based upon them, convicts them of error. We are thus working in a circle, that is a fact which there is no denying; but, as Lachmann says, the task of the critic is just this, to tread that circle deftly and warily; and that is precisely what elevates the critic’s business above mere mechanical labour. The difficulty is one which lies in the nature of the case, and is inevitable; and the only way to surmount it is just to be a critic.
The paradox is more formidable in appearance than in reality, and has plenty of analogies in daily life. In a trial or lawsuit the jury’s verdict is mainly based upon the evidence of the witnesses; but that does not prevent the jury from making up its mind, from the evidence in general, that one or more witnesses have been guilty of perjury and that their evidence is to be disregarded. It is quite possible to elicit from the general testimony of MSS. a rule of sufficient certainty to convict of falsehood their exceptional testimony, or of sufficient probability to throw doubt upon it. But that exceptional testimony must in each case be considered. It must be recognised that there are two hypotheses between which we have to decide: the question is whether the exceptions come from the author, and so break down the rule, or whether they come from the scribe, and are to be corrected by it: and in order to decide this we must keep our eyes open for any peculiarity which may happen to characterise them.
One of the forms which lack of thought has assumed in textual criticism is the tendency now prevailing, especially among some Continental scholars, to try to break down accepted rules of grammar or metre by the mere collection and enumeration of exceptions presented by the MSS. Now that can never break down a rule: the mere number of exceptions is nothing; what matters is their weight, and that can only be ascertained by classification and scrutiny. If I had noted down every example which I have met, I should now have a large collection of places in Latin MSS. where the substantive orbis, which our grammars and dictionaries declare to be masculine, has a feminine adjective attached to it. But I do not therefore propose to revise that rule of syntax, for examination would show that these examples, though numerous, have no force. Most of them are places where the sense and context show that orbis, in whatever case or number it may be, is merely a corruption of the corresponding case and number of urbs; and in the remaining places it is natural to suppose that the scribe has been influenced and confused by the great likeness of the one word to the other. Or again, read Madvig, Adu. Crit., vol. I, book i, chap. iv, where he sifts the evidence for the opinion that the aorist infinitive can be used in Greek after verbs of saying and thinking in the sense of the future infinitive or of the aorist infinitive with άν. The list of examples in the MSS. is very long indeed; but the moment you begin to sort them and examine them you are less struck by their number than by the restriction of their extent. Almost all of them are such as δέξασθαι used for δέξεσθαι, where the two forms differ by one letter only; a smaller number are such as ποιησαι for ποιήσειν, where the difference, though greater, is still slight; others are examples like ήκιστα αναγκαασθηναι for ήκισττ άν αναγκασθηναι, where again the difference is next to nothing. Now if the MSS. are right in these cases, and the Greek authors did use this construction, how are we to explain this extraordinary limitation of the use? There is no syntactical difference between the first and second aorist: why then did they use the 1st aorist so often for the future and the 2nd aorist so seldom? why did they say δέξασθαι for δέξεσθαι dozens of times and λαβειν for λήψεσθαι never? The mere asking of that question is enough to show the true state of the case. The bare fact that the aorists thus used in the MSS. are aorists of similar form to the future, while aorists of dissimilar form are not thus used, proves that the phenomenon has its cause in the copyist’s eye and not in the author’s mind, that it is not a variation in grammatical usage but an error in transcription. The number of examples is nothing; all depends upon their character; and a single example of λαβει̑ν {labein} in a future sense would have more weight than a hundred of δέξασθαι.
In particular, scribes will alter a less familiar form to a more familiar, if they see nothing to prevent them. If metre allows, or if they do not know that metre forbids, they will alter ἐλεινός to ἐλεεινός, οἰστός to ὀϊστός, nil to nihil, deprendo to deprehendo. Since metre convicts them of infidelity in some places, they forfeit the right to be trusted in any place; if we choose to trust them we are credulous, and if we build structures on our trust we are no critics. Even if metre does not convict them, reason sometimes can. Take the statement, repeatedly made in grammars and editions, that the Latins sometimes used the pluperfect for the imperfect and the perfect. They did use it for the imperfect; they used it also for the preterite or past aorist; but for the perfect they did not use it; and that is proved by the very examples of its use as perfect which are found in MSS. All those examples are of the 3rd person plural. Why? We must choose between the two following hypotheses:
(a) That the Latins used the pluperfect for the perfect in the 3rd person plural only.
(b) That they did not use the pluperfect for the perfect, and that these examples are corrupt.
If anyone adopted the former, he would have to explain what syntactical property, inviting the author to use pluperfect for perfect, is possessed by the 3rd person plural and not by the two other plural or the three singular persons: and I should like to see some one set about it.
If we adopt the latter, we must show what external feature, inviting the scribe to write pluperfect for perfect, is possessed by the 3rd person plural exclusively: and that is quite easy. The 3rd person plural is the only person in which the perfect and the pluperfect differ merely by one letter. Moreover in verse the perfect termination -ěrunt, being comparatively unfamiliar to scribes, is altered by them to the nearest familiar form with the same scansion, sometimes -erint, sometimes -erant: in Ovid’s Heroides there are four places where the best MS. gives praebuěrunt, stetěrunt, exciděrunt, expulěrunt, and the other MSS. give -erant or -erint or both. Accordingly, when the much inferior MSS. of Propertius present pluperfect for perfect in four places, fuerant once, steterant once, exciderant twice, Scaliger corrects to fuěrunt, stetěrunt, exciděrunt. Thereupon an editor of this enlightened age takes up his pen and writes as follows: “It is quite erroneous to remove the pluperfects where it can be done without great expenditure of conjectural sagacity (steterunt for steterant and the like), and not to trouble oneself about the phenomenon elsewhere.” I ask, how is it possible to trouble oneself about the phenom
enon elsewhere? It does not exist elsewhere. There is no place where the MSS. give steteram in the sense of the perfect steti, nor steteras in the sense of the perfect stetisti. Wherever they give examples of the pluperfect which cannot be removed by the change of one letter — such as pararat in i. 8. 36 or fueram in i. 12. 11 — those are examples where it has sometimes the sense of the imperfect, sometimes the preterite, but never of the perfect. And the inference is plain: the Latins did not use the pluperfect for the perfect.
Scaliger knew that in the sixteenth century: Mr. Rothstein, in the nineteenth and twentieth, does not know it; he has found a form of words to prevent him from knowing it, and he thinks himself in advance of Scaliger. It is supposed that there has been progress in the science of textual criticism, and the most frivolous pretender has learnt to talk superciliously about “the old unscientific days.” The old unscientific days are everlasting, they are here and now; they are renewed perennially by the ear which takes formulas in, and the tongue which gives them out again, and the mind which meanwhile is empty of reflexion and stuffed with self-complacency. Progress there has been, but where? In superior intellects: the rabble do not share it. Such a man as Scaliger, living in our time, would be a better critic than Scaliger was; but we shall not be better critics than Scaliger by the simple act of living in our own time. Textual criticism, like most other sciences, is an aristocratic affair, not communicable to all men, nor to most men. Not to be a textual critic is no reproach to anyone, unless he pretends to be what he is not. To be a textual critic requires aptitude for thinking and willingness to think; and though it also requires other things, those things are supplements and cannot be substitutes. Knowledge is good, method is good, but one thing beyond all others is necessary; and that is to have a head, not a pumpkin, on your shoulders and brains, not pudding, in your head.
