Augustine's Classical Readings

James J. O'Donnell

Recherches Augustiniennes 15(1980) 144-175

Cicero, it is said, once declined an invitation to speak on the ground that he had not read anything the day before.[[l]] The anecdote depicts a sober truth known to all biographers: that the history of what a subject reads and hears is potentially as useful to the reconstruction of the development of a lifetime's thought as the history Of what the subject wrote, said, and did. So it is with Augustine.

The ideal study of Augustine-the-reader would be quadripartite, faithful to the four mutually exclusive categories of books which he knew: Latin classics, non-Christian works translated from Greek (chiefly works of Platonic philosophy), patristic literature (mainly Latin, but including Greek fathers in translation), and the Bible.[[2]] But if these categories accurately reflect Augustine's own classification of the books he came into contact with, it should be possible to undertake any one of the four parts of this study independently. For reasons which I will now adumbrate, such a program is preferable.

It would be foolish, of course, to pretend that Augustine's debts to these four classes of ancient literature have not been the subject of extensive study in the past. Combes, Marrou, Testard and Hagendahl have studied his debt to the Latin classics; Theiler, Henry, Courcelle, O'Meara, and many others his reaction to Neoplatonism; Altaner above all others his use of patristic literature; and Pontet, Knauer, and La Bonnardière his use of scripture. Many of the works of these authors are deservedly regarded as magisterial. They all share one prudent feature in common, however. They are studies in Quellenforschung, that is to say, in the dogged pursuit and (often) unerring discovery of the sources that lie behind various aspects of Augustinian thought and expression. They proceed from the surviving works to a comparison with other surviving works of ancient literature, and usually confine themselves to identifying parallels between the various literary artefacts. But the information such studies produce is often controversial; there is a natural temptation to see parallels where none exist and to assume dependence where coincidence, misreport, and hazy memory are really to blame.

At one point, Pierre Courcelle identified three principal methods of Quellenforschung:[[3]] doctrinal, grammatical, and philological. The first compares points of doctrine held in common by two authors; the second looks for the residue of a source's vocabulary and syntax in the borrower's own work; while the third (Courcelle's own, a method much in favor in this century) holds that "the slightest textual parallel, in regard to sources, is infinitely more conclusive than a large number of doctrinal similarities, provided that one knows how to use this fact.[[4]] It will immediately be seen that all three methods run the risk of seeing parallels where none exist, but possess no inherent means of checking their own conclusions.

Hence I propose a fourth method, not independent of the first three, but complementary to them: the minimalist or "cynical" approach. This method insists on asking just when and under what circumstances we may conclude that a given author (in our case Augustine) actually sat down to read the work of a presumed source. By focusing on the act of reading rather than the practice of citation and demanding that the resulting calendar of readings be as plausibly simple as the evidence will allow, the "cynical" method offers the long-desired means for checking the conclusions of the other three methods and for reaching new conclusions inaccessible to the other methods. Here chronology (which is unusually well-established in Augustine's case, thanks in large part to his Retractationes) is crucial. If we find, for example, that Augustine makes a dozen incidental references to a given classical text over the 44 years of his career as Christian writer (A.D. 386-430), when are we to assume that he read the book in question? Must we assume that on all twelve occasions when the work is cited, Augustine had actually taken the time to sit down and read through the complete work in question, pausing to jot down the quotation which he happened to use in his next literary effort? Is not a likelier conclusion in such a case that he read the work carefully once, at an early age, and that memory (alone or aided by a re-check of the text in question) is responsible for the later allusions unless there is reason to think otherwise?

To answer these questions, we will also need to study the nature of memory itself. How much of Augustine's classical learning came from iterated careful study of the books themselves and how much depended on the memory he stocked assiduously during his two decades (from his teens in the 360's to his retirement from the Milan professorship in 386) as a student, then as a professor of the Latin classics? Surprisingly, this question can be more precisely answered than has been the case in the past by just the minimalist approach I suggest above. If we subtract from the total body of Augustinian classical citations those cases where we can demonstrate that he was working directly from a text he had just recently been reading carefully (these cases will include a srmall number of source texts but a high proportion of the total number of citations), we will have a much more clearly defined corpus of citations left in which the role of memory comes into question.

Finally, our study will begin from the chronological data which are so plentiful and (when compared to other ancient writers) so well-fixed in Augustine's case; but it will also throw some light on the date of some of the less well-established works. (For if we see that all the references to a particular ancient work, as in fact is the case, date from 412 and later, except for one stray reference in a work conventionally dated 406/8, but whose range of possible dates as indicated in the Retractationes can be stretched as far as 412, is it not likely that the work in question in fact comes from the later date [[5]]?)

This paper confines itself to the study of Augustine's readings in the non-Christian Latin literature of antiquity, capitalizing upon the best opportunity to implement the principles of the cynical method. In each of the other three categories listed above (translations from non-Christian Greek literature, patristic authors, the Bible), substantial uncertainty exists concerning the identity of the works available to Augustine, the precise form of Latin text available (particularly problematic in the case of Platonic translations, but even pertinent to his biblical studies), and sometimes even the contents in general of the works in question (especially in the case of Porphyry). Where doubt exists about the exact text Augustine may have read, the role of coincidence, memory, lost intermediary works, and word of mouth becomes a crucial, uncalculable factor in the development of parallels in thought and expression. But no such doubt exists for the bulk of classical Latin literature; Cicero and Varro are the chief exceptions, but they are not crippling ones, as we shall see.

The best reason of all for pursuing the study of Augustine's readings in the Latin classics first is the excellence of the scholarship on whose foundation this paper builds. The dominant work on this subject is of course Harald Hagendahl's meticulous and comprehensive Augustine and the Latin Classics[[6]]. Hagendahl's two volumes pursue separate but related purposes: to catalogue all known instances of Augustinian allusion to classical Latin literature (Volume I: Testimonia ), and to ascertain Augustine s shifting attitudes towards that literary heritage (Volume II: "Augustine's Attitude"). In the precise sense outlined above, Hagendahl's work is a study in Quellenforschung. It studies the appearance of classical erudition in Augustine's works without formally asking where and when Augustine developed that erudition. Hence there is room for this paper.

Two principal conclusions of Hagendahl should be stated at the outset with the first of which I agree, with the second of which I agree in part, but to both of which I will be able to add further documentation. Hagendahl almost broaches the question I am about to study in depth when he seeks a chronological Pattern in Augustine's citations. He notes the unavoidable: that after the early works (c. 386-391, before Augustine's ordination) there is a long dry spell for classical allusions in his theological works, until the early 410's, when Augustine began a fresh program of classical readings to fortify himself for the writing of the De civitate Dei. Hagendahl is perfectly correct in observing this pattern; I will provide detailed documentation and explore the nuances of the pattern.

I am not the first writer to express hesitation over his conclusions, however, in an important section, "Quotations from memory or from books?[[7]]" Hagendahl was certain he possessed the answer to this question: "From a psychological point of view there is not the remotest probability of Augustine's quoting throughout from memory -- even if we make allowance for an excellent memory and for the mnemonics of antiquity."[[8]] Unfortunately, it is precisely at this point that the refusal to pursue the question of Augustine's reading habits interferes with the possibility of certainty in making such a conclusion.

For one must make distinctions between different classical authors and works on the one hand, and between various types of citations on the other. No one would question that twenty lines of ad verbum citation from Cicero in the De civitate Dei came from a fresh reading of the text and transcription thereof. But if Augustine casually refers to the Terentian line, "Ne quid nimis," in the midst of a sermon on Psalm 118 (Test. 622c),[[9]] it is unlikely that he has derived this tag from a recent and careful reading of the Andria. It is my belief that if we examine Hagendahl's testimonia with care, we can make those distinctions and draw from them a provisional pattern indicating which works of classical literature Augustine read after his entry into ecclesiastical life and roughly when he read them.

That our procedure will be, at the outset, to follow Hagendahl's catalogue should provide some a priori verification of my conclusions.[[10]] For Hagendahl's purpose in assembling his catalogue was to show, within the limits of reason, how broad Augustine's familiarity with classical literature was and to include every testimonium possibly could.[[11]] If I explain the basis for all his testimonia, there will be less room to criticize my study for ignoring inconvenient evidence. I agree with Hagendahl that statistical surveys in this matter should not be thought numerically accurate, even if they offer a specious show of precision. But it is my experience in the decade since Hagendahl's survey appeared that there is not much in the way of clear and cogent evidence for Augustinian familiarity with classical Latin texts that Hagendahl has missed. I know of less than half a dozen places where I would insist on adding anything to his catalogue and less than half a dozen places where I would take the trouble to argue that his testimonia are not cogent. That this should be so is not surprising, for the reason mentioned above: that in the case of classical Latin literature, where the source texts for the most part survive and where Augustine is careful about his quotations, it is not difficult for us, in possession of much the same classical library that Augustine used, to discern citations and their sources [[12]]. After the following catalogue, in which many of my conclusions will be indicated author by author, I will outline a tripartite classification of types of reading on which such citations can be based, then conclude with a summary catalogue of what Augustine read and when. Again, at every stage in this catalogue, I assume the reader has access to Hagendahl's catalogue of testimonia to check my assertions.[[13]]

1. Apuleius:[[14]] Clearly an author whom Augustine read directly and quoted at length and ad verbum. According to the accepted chronology, Augustine knew Apuleius only by reputation in 408/9 [[15]], then appears to have come into possession of a codex of his works by 412/3, but to have studied it selectively. Thus he knows of the Apologia, but does not cite it textually; he quotes and refutes the De deo Socratis extensively in D.C.D. VIII-X (415/7); he has exactly one quotation from the De mundo in D.C.D. IV (413/5); of course he uses the pseudo-Apuleian Asclepius from the same codex extensively in D.C.D. VIII (425/7); he has a sketchy knowledge of the contents of the Metamorphoses in D.C.D. XVIII (c. 425?). The one apparent echo of the De deo Soc. in De Genesi ad litteram III, 9, 13, is probably a deliberate rectification inserted at the last moment before 'publication' (that work seems to have occupied Augustine until 414 or 415), as the form of introducing the allusion (" Non ignoro...") seems to indicate.

A pattern then emerges. Augustine came into possession of Apuleius c. 412, and read through the whole codex to see what was there. That he employed his new knowledge first in Epp. 137-38 (to Volusianus and Marcellinus) makes it very likely that the works were current in precisely the circles whence the impetus for writing D.C.D. came in the first place. This is full and sufficient explanation for the central role accorded the Apuleian texts in D.C.D. VIII-IX and further illuminates the genesis of that work's structure. Thus Augustine restudied the De deo Soc. and the Asclepius (or perhaps merely had recourse to extensive notes taken at the time of first reading) when he came to the appropriate point in the composition of D.C.D..[[16]]

Apuleius had entered the debate with Volusianus and Marcellinus initially in Ep. 136, 1 (412) when Marcellinus reported that the detractors of Christianity in his circle "Apollonium siquidem suum nobis el Apuleium aliosque magicae artis homines in medium proferunt, quorum maiora contendunt extitisse miracula". Two possible sequences of events from that point can be hypothesized:

1) Apuleius was a favored author of the local philosophical debating society which Volusianus joined upon his arrival in Africa after the sack of Rome; Marcellinus' letter gives a hint of that. Augustine then (we assume) heard more in lost letters or by word of mouth (the gap in substance between Ep. 138 and the dedicatory epistle of D.C.D. is sufficient to prove there was more communication on these subjects than we have surviving direct evidence for). Augustine then obtained a codex from these same correspondents.

2) Augustine read the mention of Apuleius; it rang a bell with something he already knew, so he sought out a copy of his work independently, to undercut his opponents in controversy by mastering one of their favorite authors. Such one-upsmanship is very frequently a factor in Augustine's parades of classical erudition, as we shall see.

It cannot be determined factually which of these hypotheses is more probable (I incline to the former), but that one of them is true seems to me virtually certain.[[17]]

2. CORNELIUS CELSUS.[[18]] Two testimonia (one a quotation, one a description of his work on all the philosophical sects in six books) from opposite ends of Augustine's career: the Soliloquia (386/7) and the De haeresibus (429).[[19]] The first quotation is merely a copybook definition of wisdom as the summum bonum, dolor corporis as the summum malum. This may have come through the Hortensius of Cicero (my guess) or through any other philosophical handbook. That he would remember the man and his work forty years later is evidence that they worked their way into Augustine's memory in a moment of striking impression. Since there is no reason to think that Celsus made that impression directly (else he would appear as the source for some of Augustine's many other minor excursions into the history of philosophy, notably in D.C.D. VIII), is it not most likely to assume that it was indeed the Hortensius that brought Celsus so memorably but trivially to Augustine's attention?

3. CICERO:[[20]] The most difficult case of all because of the depth of Augustine's familiarity with Cicero from an early age and because here a couple of crucial works have largely disappeared. I continue to follow Hagendahl for the list of testimonia, while preferring Testard's interpretation of the data on several points;[[21]] I will summarize the data for each work of Cicero.

a. General references: From the whole range of Augustine's career as writer. The only point worth observing is the passage in Ep. 9 (410, to young Dioscorus in response to his questions about Cicero -- this letter is the source of a disproportionately large number of Augustine's Cicero testimonia, not because Augustine had been reading Cicero but because Dioscorus made him talk about Cicero), where Augustine took a haughty posture of disdain towards those things which he learned in his youth, but even in that stylized disdain, there is the almost hidden concession that "aliqua ex eis in animis eorum nimia consuetudine" might still be found, even in bishops. (The plain meaning of Augustine's claim in the same passage that he did not have access even to a manuscript of Cicero might be impugned precisely on grounds of the stylization of the passage; but it stands as a tempting hint that Augustine either did not have, or did not choose to make use of, access to books of classical authors at all times in his episcopacy.) For the rest, Hagendahl's general references only show that Augustine was aware of who Cicero was, when he lived, and what were the broad outlines of his political and literary career.

b. Pro Archia. Ep. 231, 3 (429) has one anecdote about Themistocles, identical in content to one in the oration but completely different in wording. Even if the speech is the source, the allusion is not likely the result of any recent reading of the text itself.

c. Pro Caelio. This oration provides a good opportunity to classify two of the most common types of citation in Augustine. The earlier allusion comes in the De utilitate credendi (392); for there are many allusions to classical literature in the philosophical and theological works Augustine wrote in the five years or so following his conversion. (In the case of Vergil, for example, we shall see that one-fifth of all allusions in Augustine come from this period, another half coming from D.C.D., and all other works providing the remainder.) I shall not attempt to establish any strict chronology of readings for this early period, but will content myself with observing the presence of such citations. Augustine clearly did not set aside his memories or his books from his professorial career until he finally made the entry into public ecclesiastical life at Hippo in 391. One assumes that his memories were so fresh at this period that a larger number of citations than would be the case later would be strictly a matter of memory, and another large portion would be found in the books themselves at the urgent prompting of memory. There are a few cases where Augustine seems to have set himself down to confront a whole text afresh, but those occasions are strictly limited my surmise is that his examination of Cicero's Academica (see below) for the composition of his Contra Academicos is the only significant such case.

The other echoes of the Pro Caelio occur at precisely the other end of Augustine's career, in his Contra Julianum (423). Here (and in the Opus imperfectum contra Julianum [429/30]) Augustine was confronted with a controversialist who quoted the classics himself frequently and ostentatiously.[[22]] Augustine was concerned to show himself undismayed by this show of erudition and he seems to have done some very sketchy browsing in his library. In one instance, he corrects the wording of a quotation from Terence (Test. 630); once he picks up Julian's use of the trite "O tempora, O mores !" and caps it with a more recondite allusion to the Pro Caelio (Test. 80). The best reason for disbelieving in any serious review of classical literature in preparation for writing the works against Julian is precisely in the breadth and superficiality of the allusions made; one would have to imagine Augustine digesting a perfectly immense corpus of Latin literature and extracting from it only very isolated bits of verbiage. Easier to assume the use of memory and a superficial flipping of the pages -- and perhaps ancient notebooks (more on this hypothesis at the end of my section on Cicero).

d. In Catilinam I-IV. Echoes in the early works (386/91); the C. Jul. (423); D.C.D. II, III, and XIV (413/18); an undated letter; and the De consensu evangelistarum (a single line later to appear in D.C.D. -- on the date of the De cons. evang. see appendix below; the conventional date is c. 400).

e. Pro rege Deiotaro. Apart from the De beata vita (386), there is only one line cited, in the Contra Faustum (397/8), epigrammatic in form and specifically attributed by Augustine to memory (C. Faust. XXI, 16, Test. 95).

f. Divinatio in Caecilium. One memorable line appears in the treatises on the gospel of John (Zarb assigns this particular treatise to 418), together with another line from the Orator. In both passages the subject is arrogance. It is scarcely credible that Augustine turned over both of these works to turn up these two lines (see below on the Orator, which otherwise appears after 400 only in the last book of De doctrina christiana [426]). That the two lines are so apposite suggests the possibility that Augustine was working from some kind of 'secondary source,' some florilegium Ciceronianum, perhaps even a collection of extracts made by the young student of rhetoric for future adornment of his panegyrics and the like, then put to a different use by the bishop. It is notable that of the quotations from the non-philosophical works of Cicero, an immense majority fell into several trite categories: word definitions or etymologies, moralistic epigrams (as here), and rhetorically brilliant epigrams. These selections represent precisely the tastes of the rhetorical culture of the time. I hold that if Augustine ever read Cicero so superficially (so very nearly witlessly), it is more likely to have been in his youth than in the years of his episcopate.

g. Pro Ligario. A single memorable line on misericordia crops up in the Contra Adimantum (394) Ep. 104.16 and D.C.D. IX, 5 (415/17). Another memorable line occurs in Ep. 138 (412).

h. Pro Marcello. One memorable line appears in four different sermons of widely divergent dates.

i. Pro Milone. The Quaestiones in Heptateuchum (420) quotes a line from the Twelve Tables in the same words as the quotation appears in the Pro Milone.

j. Pro Quinctio. A single loose but explicit quotation appears in the De cons. evang. (date? see appendix).

k. In Verrem, actio secunda. Three echoes before 391; one from the De cons. evang. (see appendix) repeated in D.C.D. II, 27 (413); one in D.C.D. XIX, 5 (425/7?); and a single rhetorically inflated line in the Contra Cresconium(405-6 -- a work which contains several classical allusions, for Cresconius was a grammaticus himself and Augustine took the opportunity to match his erudition ostentatiously): "O scelus, o portentum in ultimas terras deportandum !" (Test. 104) (This line is rather typical of the kind cited from all the orations.)

(Philosophical Works)

1. General references occur only c. 386-91 or in the Confessions (referring to the period before 386) and Retractationes (referring to the Contra Academicos).

m. Academica posteriora (= ac. I). As one might expect, there is a burst of echoes and lengthy direct quotations in the Contra Academicos (386). A single line eulogizing Varro is quoted more than once in D.C.D. (Test. 124). There are three logical possibilities for the generation of this last citation, which are worth making explicit for purposes of showing their comparative likelihood;

a) The line may have been attached somehow to Augustine's copy of Varro's Antiquitates, a note jotted there many years before, perhaps even attached by an enterprising copyist somewhat as we employ publishers' blurbs today.

b) Augustine may have remembered Cicero's praise for Varro and spent a few minutes turning over volumes looking for the exact wording.

c) Augustine may have reread the Academica posteriora from one end to the other, picking up this line in passing. There is no other indication for this possibility. It is just barely conceivable if one imagines Augustine considering, then discarding, the idea of refuting Academic skepticism all over again in D.C.D.

Clearly, the possibilities are given here in rapidly decreasing order of likelihood; they exemplify the ease with which relatively simple explanations can be generated for many of Augustine's citations.

n. Academica priora ("Lucullus "= ac. II). Echoes from c. 386-91; passages in Ep. 104 (to Nectarius of Calama, a learned correspondent to whom Augustine always quoted the classics) and the works against Julian (but in both these cases Cicero, and even the exact words of the quotation, first came into question through the correspondent, not at Augustine's instigation). There are besides this only three small points occurring in D.C.D. VIII, XVI, and XIX (415/27), bits of unattributed philosophical and scientific lore. (Hagendahl also gives seven eases where Augustine defines man as "animal rationale, mortale " as allusions to Ac. II, 21 (Test. 126), but that is scarcely as instance of any conscious quotation in the case of so familiar a commonplace.)

o. De divinatione. In the early books of D.C.D. (413/17) Augustine showed a general awareness of the contents of this work and had one or two very small points of fact from it; this probably bespeaks a hasty review made when D.C.D. was being planned. He apparently concluded that the work was not worth detailed restudy as it was irrelevant to his argument. A single quotation in the Opus imperf. c. Jul. (429/30) occurs to illustrate an obscure word or meaning of a word; this indicates either very specialized memory or some intermediary secondary source, if only the old notebooks hypothesized above.

p. De fato. A conscious effort to provide a coherent refutation of this work is one of the main points of book V of D.C.D. Hence the work must have been at least reviewed (rather thoroughly) for the occasion. It is worth considering whether such a process of review, in the case of a reader with an ancient rhetorician's skills and memory, would not be rather different from a similar review by a modern reader. In general, it seems likely that an ancient reader would read more slow]y than we the first time through a work, but perhaps more quickly and with greater comprehension on a later occasion.

q. De finibus bonorum et malorum. Early echoes (c. 386-93); a very faint verbal echo in the De fide return invisibilium (c. 400); quotations in Ep. 118 (430 -- to Dioscorus); one vivid word picture and one footnoted citation on the alleged agreement of Platonici with Stoici in D.C.D. V and IX (413/7); and a few lines introduced by Julian in the works against him. Only in D.C.D. can there be any question of direct reexamination.

r. Hortensius. Augustine's favorite work of Cicero, by far, as his still mostly positive testimony in the Confessions would lead us to expect. Many quotations in the early works (c. 386-91); a copybook definition ("Beate vivere omnes velle") frequent at all periods; a definition of happiness in the De beata vita (386), De trin. XIII (416/9?), and Ep. 130 (411/12); several long quotations in De trin. XIV (416/9?); random lines of direct quotation in D.C.D. III, V, XXII (413/15, 425/27), and some long direct quotations in the C. Jul. (423). This work provides one of the hardest cases to judge in all this catalogue. The original text of the Hortensius does not survive for comparison, but it is clear that it is a work which impressed him deeply (and presumably had a correspondingly great claim on his memory). It is hard to avoid the conclusion that he looked at the work afresh at least once in the decade between 410 and 420; but it is impossible to determine how serious those reexaminations were, or how frequent. It is perhaps testimony to Augustine's lingering respect for the work which first inflamed him at the age of eighteen with a love of wisdom that he does not make it a particular target of his attack in D.C.D. Perhaps here, as in almost no other case of the Latin classics, one might be justified in imagining a streak of sentimental affection even in the old bishop; or perhaps he did it all with memory.

s. Laelius de amicitia. The few citations are widely scattered and scarcely merit to be considered quotations. An aphorism of Cato's makes faint appearances in Conf. IX (397/401) and Ep. 73 (404), but even there it is not likely that Augustine himself even remembered were he had first read the saying.

t. De natura deorum. This is clearly one of the works pulled out for study by Augustine in 413/15 preparatory to the writing of D.C.D.; its traces appear clearly in D.C.D. IV-V, with further echoes in books X and XXII (probably from notes made at the time of the first reading). Other quotations in Ep. 118 and the C. Jul. are by now predictable. The only surprises are the two quotations in the De cons. evang. (see appendix); one of these citations recurs in D.C.D. VII (Test. 216), while another comes from a part of De nat. deorum carefully scrutinized in D.C.D.

u. De officiis. Four definitions/etymologies, from works scattered the whole length of Augustine's career; and three epigrams, similarly scattered.

v. De re publica. There are minor echoes, mostly brought up in the first instance by opponents, in Ep. 91 and 104 (both to Nectarius) and C. Jul. (423); one allusion in the Contra litteras Petiliani (401-5) has an error which proves that it comes from memory (see Hagendahl's note, 121-22 ad Test. 250c); the great bulk of citations, often at length, occur in D.C.D. (from the whole range of that work, books II to XXII) and in Ep. 138, concerning the genesis of D.C.D. Clearly, this work was read with care and attention at the outset of Augustine's work on the magnum opus et arduum.

w. Timaeus (Cicero's translation). Frequently and carefully cited and mentioned in D.C.D., starting with book VIII (415/17). Other allusions all occur in works concerning whose date there is doubt: De cons. evang. (see appendix); De trin. IV (whole work finished 416/9, perhaps even early books emended at that time); and Sermon 241 (405/10, according to Kunzelmann). The last item deals extensively with the thought of Porphyry; hence the coincidence of quotations and allusions between that sermon and D.C.D. VIII-X does not permit resolution of the two works' relations. Sermon 241 may well be contemporary with D.C.D. VIII-X (my belief), but D.C.D. may have derived those citations from the sermon. At any rate, we can assume the Timaeus was read with care sometime shortly before it began to be used in D.C.D. I will tentatively assign the reading to 412/13.

x. Tusculan Disputations. Apart from one epigram cited in the Confessions (and repeated in D.C.D. IV), there are no testimonia which can be conclusively dated after 391 and before 410/12. The borrowings in D.C.D. (they occur in books I, V, VIII, IX, XIV, and XXII) are not as extensive and serious as the ones from works like De re publica and De fato, and may indicate a slightly less attentive reading. Much of what appears from this work is of the nature of definition and doxographical information. Once again, it is good to note, a definition occurs in D.C.D. VIII which had already (?) occurred in De cons. evang. in slightly fuller form; see appendix.

(Rhetorical Treatises)

y. General references are of no import.

z. De inventione. Apart from early reminiscences and a lengthy quotation defining virtus taken over in the De diversis quaestionibus LXXXIII (388-95 -- we may associate this with the 'early' citations for the moment), there are only a handful of borrowed definitions: De doct. christ. (396/426); De trin. XIV (416/9 ?); Sermon 150 (413/14); and C. Jul. (423)- In the rhetorical works we should be more hesitant than ever in assuming direct dependence, since so many definitions and bits of expression would have been so thoroughly ingrained in Augustine in his early years.

aa. De oratore. Much the same pattern of definition/citation occurs. A single epigram pursues a wandering course from the C. A cad. (386) to the C. Cresc. (405-6) to D.C.D. IX (415/17), certainly passing the intervening years lodged in memory rather than in a library.

bb. Orator. Some definitions appear in De doct. christ. IV (426); bits of fluff and definition show up elsewhere, most memorably the description, magnum opus et arduum, which Augustine applied first to his De doct. christ., then to D.C.D.

(Letters)

cc. All that remains of all of Cicero's letters in Augustine's works are three echoes; the only two datable ones come from 427 and 429. No significance should be attached to these.

dd. Fragments unplaced. In perfectly predictable places: Ep. 143 (412 -- when preparing D.C.D.); D.C.D. itself; Ep. 143 (to Nectarius); C. Jul. (423) in the latter case both quotations originally being made by Julian himself.

It is worth pausing briefly to summarize what we have learned of Augustine's contacts with Cicero. It seems probable that whatever deficiencies we insist on imparting to memory were more likely to have been made up through the medium of some lost intermediate 'secondary' source, such as the home-made florilegium hypothesized above. The triviality of the allusions and their wide distribution over both Cicero's works and Augustine's makes it altogether unlikely that Augustine pursued any course of serious readings in Cicero other than the limited one he undertook in preparation for De civitate Dei. From that preparation, only six works emerge as having been restudied; I will give them in descending order of the importance Augustine attributes to them in his work (and probably of the attentiveness and effort which he devoted to their reexamination): De re publica, De fato, Timaeus, Tusculan Disputations, De finibus bonorum et malorum, and De natura deorum. Did Augustine ever seriously reread any of his Cicero other than those works on that occasion ? Only my suspicion of occasional reference to the Hortensius and the possibility of some sketchy dipping into the rhetorical works in 426 when De doctrina christiana was being finished can give any indication of that. I would certainly side with Westard against Hagendahl in assuming that the speeches went unread after about 391 (though copies of them were probably available for verifying references).

4. CLAUDIAN:[[23]] It is notorious that three lines of Claudian, abridged (to leave out mention of a pagan god, some suppose) to two by Augustine, appear in D.C.D. V (415) and are picked up later at Orosius, VII, 35, 21, similarly abridged. The tantalizing feature of these lines is that they were not written until 395, so Augustine must have encountered them for the first time after that date. I am not inclined, however, to believe Augustine was reading Claudian directly in extenso, if only because he would have quoted the three lines intact, precisely because he was attempting to underline the paganism of Claudian. The quotation seems rather to have come to Africa as part of a package of information on events of 394 (and what some Christians of the time, and the more credulous of modern scholars, believed to be the ' pagan revival ' associated with the usurpation of Eugenius);[[24]] this information may very well have come to Augustine's attention in discussing the pagan- Christian issues made relevant after the sack of Rome in 410 with some of the cultured Italian refugees whose perplexities were the impetus for writing D.C.D. It is worth noting that the passage recurs in Orosius in precisely the same isolation in an account of the events of 394. If Orosius had had more of Claudian at hand, he certainly could (and would) have used him.

5. EUTROPIUS:[[25]] Only appears in D.C.D. III-V (413/15), with sufficiently precise verbal parallels in at least one of the testimonia to indicate that Augustine consulted the work, directly but perhaps intermittently, while working on D.C.D.

6. FLORUS:[[26]] Only appears in D.C.D. III (413), with virtually direct quotation at D.C.D. III, 19 (Test. 393). Augustine's study of history manifests itself only at this narrow period (see my remarks on Livy below).

7. AULUS GELLIUS:[[27]] One anecdote about a hypocritical philosopher is recounted at D.C.D. IX, 4 (415/17) and rehashed at Quaest. Hept. I, 30 (420). Hagendahl makes much of the remark in the latter case (apparently dictated as an aside to a scribe but cropping up in our final text) to the effect that the anecdote is to be checked against the original: "Sed considerandum est quemadmodum hoc dicat A. Gellius et diligenter inserendum. " This is still not evidence, however, that Augustine had read Gellius at any length, or even that he possessed a complete copy of his work; where one anecdote survives from so vast and entertaining a collection, one is justified, I think, in insisting on the assumption of some lost intermediary, which can even have been the kind of florilegium which quoted authors' words verbatim and with which Augustine may have meant his scribe to compare his dictation. Even in D.C.D. the quotation was not verbatim, but very loose.

8. HORACE:[[28]] Horace provides the best example of Augustine's use of the classics of Latin poetry. Numerous citations appear in the De musica, first of all, derived from the handbook of Terentianus Maurus; these need not detain us. The remaining echoes fall into a vague pattern:

(1) Early: De quantitate animae (387/88);

(2) 412/413 and shortly after (at work on D.C.D.): Ep. 143 (to Marcellinus), D.C.D. I and V (the latter citation being repeated in Ep. 231 [429]), Ep. 167 (415), this last repeated in the C. Jul. (423);

(3) 420 and shortly after: Quaest. Hept. (420), for grammatical examples; Contra .Mendacium (422); and C. Jul. (423),

Apart from these three groups of citations there is only the echo, surely due entirely to memory, of the phrase, dimidium animae meae, that occurs in the Confessions (397/401).

It is difficult to avoid the conclusion that Augustine was at least glancing through his copy of Horace at about the time he began work on the D.C.D. Had the poet's work subsisted entirely in memory, one might have expected to see more traces in letters to people like Nectarius and Jerome in the 400's. But the overall bulk of the passages cited is very small: a poor harvest for any diligent cultivation of the author's poetry. We are left undecided between an unusually tenacious memory that still produced ideas after being ignored for two decades or an unusually superficial skimming of the text as part of the preparations for D.C.D.

9. JUSTINUS:[[29]] Directly quoted in D.C.D. IV, further close parallels appear in D.C.D. XVIII-XXI, all concerning what one might term 'pre- Roman' secular history (Babylonian empires and the like). Apparently this author formed part of the object of Augustine's historical researches c. 412/13, with some of the information derived then being saved for use in the later books of D.C.D.

10. JUVENAL:[[30]] Hagendahl catalogues only two echoes, but I can add a third:

Test. 426: D.C.D. II, 23 (413), echoing half a line of the first satire, concerning Marius.

Test. 427: Ep. 138 (412), quoting verbatim nine lines of the sixth satire, contrasting the virtues of the republic with the vices of the empire.

Additional observation: D.C.D. I, 2 (413): , "Quo modo ergo colebatur, ut patrim custodiret et cives, quae non valuit custodire custodes ?" (Cf. Juvenal 6, 347-48 or 6,0 31-32)

I venture to suggest that if these echoes do not come from memory (verified in the second instance by a check of the exact text), then they might be attributed to an abortive intent to use the satirist for material for D.C.D. Note that the testimonia give warrant for attributing certain knowledge of Juvenal to Augustine only in the cases of the first and sixth satires, two of the most memorable and most read. It may simply have been want of time that kept Augustine from looking further for material; or it may have been the realization that the theme of the first books of D.C.D. was that affairs at Rome were as bad before the birth of Christ as they were in the early fifth century, and that Juvenal gave evidence for a period that strictly fell after the entry of Christianity into the world. (The contents of the one verbatim quotation indicate that this was indeed Augustine's purpose.) Augustine may then have avoided Juvenal to avoid providing an opportunity for nit-picking by his critics.

11. LABEO:[[31]] Quoted only in D.C.D., in books II and XXII, evincing awareness of only four passages in Labeo.[[32]] Augustine may have looked at his work in 413, but the sketchiness of the use combined with the nonsurvival of Labeo's works makes some intermediary source more likely.

12. LIVY:[[33]] There is one possible echo in Ep. 104 (to Nectarius, of course), but for the rest the testimonia are exclusively from D.C.D. I-V. Even where the appropriate books of Livy survive, it is sometimes hard to know whether Augustine knew them directly, or only through periochae.

It is appropriate to remark at this point, having catalogued the use Augustine made of Eutropius, Florus, Justinus, and now Livy, on Augustine's general practice in using historical writers. First of all, it is striking that they appear echoed in his work for all intents and purposes exclusively in D. C.D. Augustine does not seem to have read Roman or secular history with much attention or interest, until it was forced on him as a subject in the wake of the sack of Rome. His disdain for the subject can be seen vividly in several places. In De doctrina christiana II, 28, 42-44 he thinks secular history occasionally useful to explicate the sacred books (but Roman history as such would not have much role there). In D.C.D. III, 18, 11-12 (CCSL ed.) he desists from narrating the whole history of the Punic Wars with a barely concealed sneer: "Si enarrare vel commemorare conemur, nihil aliud quam scriptores nos erimus historiae." (Postponing historiae to the last place in the sentence is a sly touch.) We know that he commissioned Orosius, the headstrong young Spaniard, to do the dirty work of historical research for him (even if, of course, he was to be sharply disappointed by the witlessness of the results).

For our immediate purposes, then, we can conclude that Augustine did some rummaging in volumes of Roman history c. 413; the exact sequence and substance of his readings has been a subject of controversy and can probably never be determined with great certainty. What is clear is that Augustine was only (to use an Augustinian distinction) using the authors as reference works, rather than enjoying them as literary creations. Sallust (v. infra) is the only historian who interested him other than as a random source of polemically useful facts.

13. LUCAN:[[34]] One early quotation (De Genesi contra Manichaeos) recurs at De Genesi ad litteram III, 2 and D.C.D. XV--a good example of what we might term 'internal transmission' from one work of Augustine's to another, when the same subject and the same pertinence for the quotation recurs. Apart from that the bulk of citations are in D.C.D. and contemporary works (and three in C. Jul.). The only exceptions are a very faint echo of the Cato passage in Lucan II appearing in the De bono coniugali (401) and a verbally explicit quotation concerning the Jews in the De cons. evang. (date ? see appendix). Like Horace, Lucan is an author who came to mind in the early 410's; whether a copy of his works came to hand for more than the verification of references is a subject that cannot be ascertained with certainty. My concluding discussion below will deal with this recurring issue.

14. LUCRETIUS:[[35]] It is with surprise that one discovers that Lucretius, though known to Augustine, seems to have had little claim on his attention. There are early citations (De utilitate credendi, Contra Academicos, De libero arbitrio); echoes in the Confessions (397/401),[[36]] De Genesi ad litteram (401/15), possibly in De trinitate IV (399/419), and (I would suggest, unnoticed by Hagendahl) in D.C.D. XXII, 24.[[37]]

I say 'with surprise' advisedly. In two major regards, there are important parallels between Augustine's work and Lucretius's. First, Augustine's theory of the mechanics of creation in the De Genesi ad litteram (where a couple of echoes can indeed be found), particularly the doctrine of the rationes seminales, has a strikingly Lucretian sound to it; second, the whole thrust of the De civitate Dei, the refutation and mockery of the claims of Roman civil religion, is precisely parallel to what Lucretius undertook in his poem (There is even a structural parallel, in that Lucretius adorned his argument with a lengthy narrative in his fifth book of the whole history of mankind, to explain how things got the way they are; books XI- XXII of D.C.D. have much the same function.)

If Augustine knew Lucretius, as it would seem he did, why did he not use more of his material for his own purposes ? First, perhaps, because of the virtually universal contempt for Epicureanism in late antiquity (by pagans and Christians alike), hence a disdain on Augustine's part to mention so disreputable a school in any favorable way; second, and in a way related to the first obstacle, when Augustine quotes classical authors in D.C.D. he is usually employing them to undermine the positions of his non-Christian opponents --e.g., quoting Sallust (not some atheistic, Christian bishop) to claim that Roman society began to deteriorate with the fall of Carthage; quoting Cicero's definition to prove that Rome was never really a res publica. The essential quality of such quotations is precisely the unavoidable, undeniable degree of respect in which the authors quoted were held by Augustine's opponents. If he tried to quote Lucretius in that way, he would be liable to the retort, "Yes, but he is an Epicurean and hence contemptible, so you need not expect us to take what he says seriously. " Thus it is that Augustine's modern readers are left with a sentimental curiosity to know what Augustine might have made of Lucretius had he dared follow more closely in his footsteps; but it would not have served Augustine's purpose to do so.

15. Nigidius Figulus:[[38]] single aetiological copybook anecdote, source unknown, in which we learn how Nigidius got his cognomen, occurs in D.C.D.V. Like the passage from Aulus Gellius, this anecdote indicates that Augustine in the early 410's had access to material about the history of ancient philosophy through some secondary source, be it florilegium or doxographical handbook, perhaps lingering in his library from his own years as a student of philosophy.

16. OVID:[[39]] The story of Pyramus is mentioned in the De ordine, because Augustine's poetaster student Licentius was writing about the story himself; apart from that there is only one suggested echo in D.C.D., and that one is so faint as to be explained as easily by coincidence and the commonplace character of the notion.

17. PERSIUS:[[40]] Only one testimonium is evidence for knowledge of more than isolated lines of Persius, a seven-line quotation in D.C.D. II. Other single lines occur (chronologically) in the De magistro (389), De baptismo (401), Ep. 118 (410 -- to Dioscorus), Ep. 132 (412 -- to Volusianus), D.C.D. II (413), C. Jul. (423 -- only at Julian's instigation), and Ep. 231 (429). There is a faint (and to my mind unlikely) echo in Conf. Once again, Augustine may have briefly handled a copy of Persius c. 412; the rest is memory.

18. PLINY THE ELDER:[[41]] Apparently a research source for D.C.D. XVXXI (418/27 ?), though the use of Solinus as well (v. infra) makes it necessary to admit the possibility of a lost intermediate source of scraps of information on natural history.

19. PUBLILIUS SYRUS:[[42]] One unexplained probable echo rings in D.C.D. IV, but the line (" Fortuna vitrea est: tum cum splendet, frangitur ") may have been proverbial and the echo is distant.

20. QUINTILIAN:[[43]] Hagendahl offers four testimonia, one of which is convincing evidence Augustine ever knew Quintilian directly; the information in these four testimonia probably remained in Augustine's mind from his years as rhetorician. The testimonia include a common anecdote about Demosthenes on oratory; a definition of antitheton; and the topic of a rhetorical declamatio. These last topics, in particular, seem to have .haunted Latin rhetorical education like recurring bad dreams, becoming detached from their origins as completely as they were detached from reality.

21. SALLUST:[[44]] We know that Sallust was a principal object of Augustine's attention while working on the De civitate Dei. Apart from quotations and allusions during the decade 410/20, what is there ? Very little: three echoes in Conf., an isolated reference in Elk. 82 (to Jerome), one in Ep. 104 (to Nectarius), and one in Sermon Denis XIV (Kunzelmann places the date before 400); and the usual clutch of references in works of c. 386-91.

Note well, however, that the Historiae of Sallust can be traced only in D.C.D., and even there often seems overshadowed by the Catilina. This confirms my easier observation (supra, s.v. Livy) about Augustine's disinclination toward the study of history at any other time in his career. Sallust, of course, was a central school-author, well-known to Augustine (at least for the Catilina) from his earliest years, and favored by him for his moral interpretation of Roman history presented in the Catilina. It is that authority which Augustine exalts in the early books of D.C.D. That he barely uses the Historiae even after taking time to consult it c. 412 is evidence of the limitations on time and energy which constrained his research at that time. As elsewhere, we see here evidence of a few texts read carefully then, but a larger number only dipped into hastily.

22. SENECA RHETOR:[[45]] The one echo (in D.C.D.) is most likely a semi-proverb half-remembered from the rhetorician's classroom. One cannot seriously imagine Augustine, in his busy years as bishop, deliberately sitting down to read the eider Seneca's drivel.

23. SENECA PHILOSOPHUS:[[46]] The most substantial presence of Seneca is in the solid chunks of De superstitione quoted in D.C.D. VI (415/17). In addition there are five lines on Fate in D.C.D. V (415) from Ep. 107 (including the famous line, Ducunt volentem rata, nolentem trahunt). If we go so far as to imagine Augustine reading the whole of Ep. 107 (hardly necessary in my estimation), there is still nothing to force us to believe he read any more of the philosopher's letters than that. The passage was sufficiently famous and pertinent to later debates to attract even a bishop's attention.

More puzzling and intriguing are the two clear echoes which survive of Seneca's tragedies, in direct quotation, but without attribution: one occurs in the Contra Faustum (397/98), the other in Sermon Frangipane VIII (24 June 401: Kunzelmann). They comprise barely two and one-half lines, from two different plays. To me even the temporal juxtaposition of the citations is still insufficient evidence to indicate that Augustine had been reading bombastic poetry in his spare time; some coincidences, after all, are only that: coincidence.

24. SOLINUS:[[47]] Used for ten natural wonders described in D.C.D. XXI; like Pliny the eider (v. my comments supra), he occurs only as a reference source superficially consulted late in D.C.D.

25. TERENTIANUS MAURUS:[[48]] A large collection of borrowings in De musica (387-90); the name alone appears in De utilitate credendi (392); and a single-line eulogy of Varro survives in D.C.D. VI (see my comments on a parallel passage in the Academica posteriora of Cicero supra).

26. TERENCE:[[49]] One of the four central authors of late Roman education (the so-called quadriga Messii: Cicero, Sallust, Terence, and Vergil), Terence provides an opportunity for a detailed catalogue of echoes designed to demonstrate my contention that Augustine is likelier to have drawn much of his later fragmentary erudition from memory than from active rereading of books. I will here catalogue every explicit reference to any work of Terence given by Hagendahl, but will rearrange them chronologically under each work. As the reader peruses these lists, he or she should ask, "Is it likely that at each of the widely scattered dates at which Augustine showed knowledge of a line or two from this play he had actually been recently rereading the work in question?" I hold it certain that the wide chronological distribution of testimonia is evidence for reliance on memory in the first instance, with reference to books only as a means of verification and confirmation.

(a) Adelphoe: 395 (Ep. 26); 412/13 (Ep. 138); 417 (Ep. 185); 425/27 (D.C.D. XIX);

(b) Andria: 386 (De ordine, De beata vita);[[50]] 387/89 (De mor. eccl.); 389 (De mag.); 391 (Ep. 21); 396 (De doct. christ. II); 397/401 (Conf.) repeated in 405 (Ep. 82); 416/19 (De trin. XIII); 417/20 (D.C.D. XIV); 420 (Quaest. Hept.); and 429 (Opus imperf. c. Jul.);

(c) Eunuchus: 386 (De beata vita, De ordine; 397/401 (Conf.) repeated in 409 (Ep. 91) and in 413 (D.C.D. II): 423 (C. Jul., twice); 425/27 (D.C.D. XIX);

(d) Hautontimorumenos: 401/5 (C. litt. Petil.); 413/14 (Ep. 155) repeated in 421 (C. Jul.);

(e) Phormio: 386 (De ordine); 397 (Ep. 40); 397/401 (Conf.) repeated in 429 (Opus imperf. c. Jul.); 413/14 (Ep. 155).

27. VARRO:[[51]] Of 150 testimonia, all but nine come from D.C.D. The nine are:

(1) From De ordine (386);

(2) Again front De ordine (386);

(3) From De quantitate animae (387/88);

(4) From De doct. christ. II (396) -- all of the first four are probably indebted to Varro's Libri disciplinarum, Augustine's model for his own abortive project to write about the liberal arts;

(5) -- (9) Prom De consensu evangelistarum, where the only direct item front Varro is the claim that the Jews also worshipped Jupiter. See appendix for date of this work.

It should be noted that there are two groups of works of Varro utilized in D.C.D., at different dates:

First group: Antiquitates rerum divinarum, books I, XIV, XV, XVI (though Augustine has at least seen the disposition of subjects for the whole of the Ant.), used from D.C.D. II onwards; and the De cultu deorum (used in D.C.D. VII only). These works were studied carefully c. 413/15.

Second group: De genre populi Romani (used in D.C.D. XVIII); De philosophia (D.C.D. XIX). These works were either studied carefully along with the others (413/15) and the notes saved for use later (c. 425/27 ?); or the works themselves were postponed for serious study during the later period.

28. VERGIL:[[52]] More testimonia in Hagendahl than for any other author except Cicero. By my count, Hagendahl gives 251 occasions on which Augustine quoted or alluded to approximately 156 specific passages of Vergil. Of those 251 occasions,[[53]] occur in the early years, while Augustine was still mainly an ex-professor of Latin; 115 occur in D.C.D. for polemical effect. Only 83 occur in all the rest of Augustine's works, spread unevenly over his entire career. The other citations tend to bunch up in the 410's, but no decade is without representatives (the De div. quaest. LXXXIII and the Exp. incho. in Rom. for the 390's, the C. Cresc. and various letters in the 400's; Ench. and De cura mort. get. for the 420's). As I will defend in detail later, I hold that the overwhelming number of these citations (and virtually all those not in D.C.D.) can be attributed to memory.

To summarize this catalogue, and to take the game of counting and weighing one (perhaps permissible) step further, the following table of numbers of pages may be useful:

Begin with the pages in Hagendahl's catalogue--359
Remove pages of Apuleius --16
-- and of Varro -- 51
This leaves -- 292

Remove the authors of the quadriga Messii:
Cicero --134
Sallust -- 19
Terence -- 10
Vergil -- 59
This leaves -- 70 pages

Seventy pages of testimonia from all the other authors of classical antiquity, equaling a little less than one-fifth the bulk of the whole catalogue; this should be a salutary reminder of the limitations on Augustine's parade of classical erudition, as we turn to the questions of what books he read; when he read them; how carefully he read them; and what books he only remembered.

It is not easy to provide clear-cut categories for the different ways in which Augustine may be concluded to have read the classics during his years at Hippo. I will employ the following threefold distinction, but only while remaining fully aware of the imprecision involved in doing so. The three suggested categories are contiguous, even overlapping, and doubt frequently can be raised to which of two a particular case belongs. Nevertheless, the schema will prove useful. What must still be borne in mind is that there is a difference between Augustine's citations of classical literature (the use to which he put his erudition) and his readings in that literature (the method by which he developed and enhanced that erudition).

1. Verification of a reference owed in the first place to memory. [This category includes both instances where Augustine was merely checking the accuracy of a quotation securely lodged in his memory and those cases where he sought out the precise form. of a quotation half-remembered already.]

2. Hasty survey or review of a text, dipping into it here and there for whatever useful material may be found. [Similarly various: this kind of reading comes closest to the first category when it consists of indolent page-flipping, looking for the memorable and important phrase; it abuts on the third category when it entails the careful reading of scattered small sections of a large work.]

3. Serious consecutive reading of an entire work or large portions thereof.

Hagendahl would readily admit the possibility of all three categories; but he would minimize the importance of the first two. Before presenting my own summary catalogue of Augustine's readings classified according to this schema, therefore, I will discuss the question of 'class one' readings in particular, i.e., verifications of remembered passages, in greater detail.

As quoted above, Hagendahl claimed that it was out of the question on psychological grounds to argue that Augustine subsisted all his life on the strength of memory53. We have seen throughout this study that this is certainly a true statement if we consider the entire range and bulk of Hagendahl's catalogue of testimonia. But if we extract from that catalogue the relatively limited number of works which we have indicated Augustine was reading carefully, how much is left to be attributed to memory ? Only if we answer that question can we assess accurately the likelihood of memory's powers being equal to the task.

One line of approach continues that used in my catalogue above to present the evidence for Augustine's use of Terence: showing the chronological diversity of testimonia to the same work or portion thereof. The greatest variety of possible loci for comparison comes in Augustine's use of Cicero and Vergil, the two most frequently quoted Latin authors. In those cases, we have the opportunity to examine the number of cases where a single passage of the classical author is echoed more than once in more than one work of Augustine's.[[54]] Furthermore, I have subjectively eliminated from this catalogue every passage which seemed to me to be literally commonplace in the ancient sense. Only half a dozen Vergil testimonia survive such relentless culling, but they are enough to be indicative:

Test. 822: Aen. i.279-285: D.C.D. III (413)