[[1.]] CASSIODORUS, Variae , XI, Praef. 8.

[[2.]] We assume that Augustine's readings of (Greek works in the original were few and comparatively insignificant. On this point, H.-I. Marrou, Saint Augustin et la fin de la culture antique (Paris4, 1958), 27-37, 631-637, requires only minor modification.

[[3.]] P. COURCELLE, Late Latin Writers and Their Greek Sources (Cambridge, Mass., 1969), 5-7.

[[4.]] Op. cit., 6.

[[5.]] Ep. 102, on which see below at note 16.

[[6.]] Göteborg, 1967 [Studia Graeca et Latina Gothoburgensia, XX]. Hereafter simply: Hagendahl.

[[7.]] Op. cit., 702-713. MARROU, Gnomon 41 (1969) 284, expressed hesitation in accepting Hagendahl's conclusions on the limited power of memory in generating citations; A. STRENNA, Revue des études latines 45 (1967) 183-84, would agree with Marrou. A. STRENNA, op. cit., 191, also observed the broad chronological pattern of Augustine's use of classical erudition, without going further to seek the pattern of readings on which that use was based, G. MADEC, Revue des études augustiniennes 14 (1968) 231, observed a chronological pattern with similar limitations. M. TESTARD, Revue des études augustiniennes 14 (1968) 47-67, already questioned some of Hagendahl's conclusions with particular regard to Cicero.

[[8.]] HAGENDAHL, op. cit., 709-710.

[[9.]] For convenience, I will thus frequently use the reference numbers assigned to individual testimonia by Hagendahl in his catalogue as a shorthand way of referring both to the original sources and to Augustine's citations of them. I am happy that this compendious form of reference will encourage readers to check my conclusions against the original data catalogued by Hagendahl.

[[10.]] Hagendahl's catalogue contains a substantial number of authors for whom there is not evidence of direct contact in Augustine's writings (e.g. poets cited here and there through the intermediacy of Terentianus Maurus). I omit such authors from my catalogue as irrelevant to the purposes of this study. For clarity, I will now give a list of those authors, using the forms of names appearing in Hagendahl: Annianus, Asper, Caecilius, Catullus, Cornutus, 'Declamatio anonymi ', Donatus, Ennius, Marius Victorinns, Naevius, Pacuvius, Petronius, Plautus, ' Poetae incogniti ', Pomponius Bononiensis, Pomponius Secundus, Pompeius Trogus, Valerius Maximus, and Valerius Soranus.

[[11.]] To be sure, Hagendahl exercised prudence in his selection of testimonia, often omitting unduly far-fetched parallels spotted by over-eager earlier researchers. This combination of breadth and prudence makes his work the ideal foundation for my own study.

[[12.]] ' Citation ' will be used in this paper as a catchall term to cover all the kinds of phenomena (from verbal echo to verbatim quotation) catalogued by Hagendahl.

[[13.]] For convenience, I have generally accepted the dates assigned to Augustine's works by S. ZARB, Chronologia operum S. Augustini (Rome, 1934), reprinting a series of articles originally appearing in Angelicum 10 (1933) 50-110; 261-285; 359-396; 478-512; 11 (1934) 78-91; and for the sermons, the dates of A. KUNZELMANN, Miscellanea Agostiniana 2 (1931) 417-520. Significant departures from their conclusions will be noticed explicitly.

[[14.]] HAGENDAHL, 17-33, including the entry (29-33) on "pseudo-Apuleius".

[[15.]] Our argument concerning Augustine's knowledge of Apuleius is valid however one chooses to date Ep. 102, even with ZARB, Angelicum 10 (1933) 296-97, to A. D, 408 or 409. But Zarb's only reason for dating the work thus is its priority to the De peccatorum meritis, which he can date no more closely than to 411/12. But the form of the citation of Apuleius in Ep. 102 is very close to the form used by Marcellinus at Ep. 136. 1 (412). It is probable, therefore, that Ep. 102 should be related to early 412 and should be read more closely as part of Augustine's anti-pagan polemics of the years immediately following the sack of Rome. (GOLDBACHER [CSEL , vol. 58, indices] dates Ep. 102 more loosely to 406/12, solely on the order of the Retractationes , and hence offers no contradiction to my suggestion. )

[[16.]] It is the merit of J.-.C GUY, Unité et structure logique dans la "Cite de Dieu" de saint Augustin (Paris, 1961), to have shown conclusively that this kind of detailed planning years in advance was unquestionably within Augustine's power. I hope at a later date to pursue the line of research opened by Guy in the belief that D.C.D. is a work of great rhetorical sophistication and one profoundly well-organized, often even in minute detail.

[[17.]] M. MOREAU, Le Dossier Marcellinus dans la Correspondance de saint Augustin (Paris, 1973), does not comment on the role of Apuleius in this debate. (One cannot rule out the possibility that Augustine had read Apuleius at some time in this philosophical youth; I am merely contending that his approach to this author in the 410's has all the signs of a fresh beginning.)

[[18.]] HAGENDAHL, 34-35.

[[19.]] COURCELLE, Late Latin Writers 192-94, argues that the passage in the De haer. actually refers to a Latin translation of a lost work (known from the Suda) by one Celsinus of Cappadocia. The question is moot and irrelevant to our present purposes. Whoever was the author, there is no evidence that Augustine ever possessed (Courcelle errs in his interpretation on this point) or read such a work at any time in his episcopal career.

[[20.]] HAGENDAHL, 35-169.

[[21.]] M. TESTARD, Saint Augustin et Cicéron (Paris, 1958); see also his remarks on Hagendahl in Revue des études augustiniennes 14 (1968) 47-67.

[[22.]] A separate study of the pattern of classical allusions in Julian's works is one of the many tasks that need doing if we are to understand this distant figure more accurately.

[[23.]] HAGENDAHL, 169-70.

[[24.]] See my "The Career of Virius Nicomachus Flavianus," Phoenix 32 (1978) 129-143; and on the general problem of 'pagan revivals,' my "The Demise of Paganism," Traditio 35(1977) 44-88.

[[25.]] HAGENDAHL, 172-76.

[[26.]] HAGENDAHL, 176-79.

[[27.]] HAGENDAHL, 179-84

[[28.]] HAGENDAHL, 184-90.

[[29.]] HAGENDAHL, 190-92.

[[30.]] HAGENDAHL, 193.

[[31.]] HAGENDAHL, 194-95.

[[32.]] Viz., distinguishing gods as good and evil (e.g., D.C.D. II, 11); naming Plato a semideus (D.C.D. II, 14); identifying demons with angels (D.C.D. IX, 19); narrating an anecdote on reincarnation (D.C.D. XXII, 28).

[[33.]] HAGENDAHL, 195-206.

[[34.]] HAGENDAHL, 206-10.

[[35.]] HAGENDAHL, 211-12.

[[36.]] Not in Hagendahl; see F. PIZZOLATO, Revue des études augustinennes 17 (1971) 55-57.

[[37.]] The influence of Cicero in this famous word-picture has already been noticed by M. TESTARD, Augustinus Magister I (1954) 193-200. Note the further parallel: D.C.D. XXII, 24, 185-88 (CCSL ed.): Quam porto delectabiliter spectatur etiam quandocumque turbatur, et fit inde maior suavitas, quia sic demulcet intuentem, ut non iactet et quatiat navigantem !

LUCRETIUS, De rerum natura II, 1-2:

Suave, mari magno turbantibus aequora
ventis, e terra magnum alterius spectare

[[38.]] HAGENDAHL, 213.

[[39.]] HAGENDAHL, 213-14.

[[40.]] HAGENDAHL, 215-18.

[[41.]] HAGENDAHL, 219-22.

[[42.]] HAGENDAHL, 224.

[[43.]] HAGENDAHL, 224-25.

[[44.]] HAGENDAHL, 225-44.

[[45.]] HAGENDAHL, 244.

[[46.]] HAGENDAHL, 245-49.

[[47.]] HAGENDAHL, 249-52.

[[48.]] HAGENDAHL, 252-54.

[[49.]] HAGENDAHL, 254-64

[[50.]] One of the echoes in De beata vita recurs in De doct. christ. II (396) and En. Ps. 118 (418); the other echo in De beata vita recurs in De trin. XIII (416/19) and D.C.D. XIV (417/20).

[[51.]] HAGENDAHL, 265-316.

[[52.]] HAGENDAHL, 316-75.

[[53.]] HAGENDAHL, 709-10.

[[54.]] Hence I exclude cases where the same line is iterated half a dozen times in a single book of D.C.D., a common rhetorical practice for Augustine.

[[55.]] As throughout this paper, I have not thought it my concern to disentangle memory from autopsy for the earliest works, essentially those of 386-91, when memory was still fresh and the willingness to spend time on the classics greater than it became when ecclesiastical responsibilities began to preoccupy Augustine's attention.

[[56.]] Note well how the list contains only two prose authors: Sallust and Cicero, both part of the quadriga Messii from Augustine's schooldays.

[[57.]] Approximately 80% of these lines come from the four authors of the quadriga (393 of the 488 lines).

[[58.]] De anima et eius origine IV, 7, 9; the man could do the same with orations of Cicero.

[[59.]] It is in that context that Augustine's famous remarks to Dioscorus (Ep. 118.9) should be read: "episcopis exponenda ea mittere cogaris: quasi vero episcopi isti, etiamsi adolescentes, codera quo tu raperis animi ardore, vel potius errore, quasi aliquid magnum haec discere curarunt, usque ad canos episcopales, et usque ad cathedras ecclesiasticas, ea sibi in memoria durare paterentur..."

[[60.]] ZARB, Angelicum 10 (1933) 488-90.

[[61.]] Bibliothèque Augustinienne, vol. 12 (Paris, 1950), p. 580.

[[62.]] CSEL 43 (1904).

[[63.]] COURCELLE, Late Latin Writers , 173-75, uses De cons. evang. I. 23, 35 to establish a connection between Plotinus, Enn. V, 1, 4-7, and Conf. IX, 10, 24. Unfortunately this connection depends on dating De cons. evang. to 398 (though on p. 176 Courcelle switches the date to 400) and hence on abandoning the one firm datum for the date of De cons. evang., the entry in the Retractationes . In any event, the alleged connection is remote and tenuous in the extreme -- too feeble to mean anything.

[[64.]] I am grateful to G. MADEC for reference to his article, "Tempora Christiana" , Scientia Augustiniana: Festschrift Adolar Zumkeller (Würzburg 1975), 118, n. 34. on the date of De cons. evang. G. Madec remains unconvinced of any need to postpone the date of completion of that work much past 404. I am not convinced that the parallels I have suggested in citation patterns between De cons. evang. and De civ. Dei can be explained away on grounds of similarity of subject. If a definite answer to this problem is to be found, it lies in the careful study of the pattern of apologetic works as a whole (including letters and sermons) which Augustine produced in the years immediately after 410. Astonishingly, no such thorough study exists, so readily have we assumed that the circumstances of the composition of De civ. Dei are self-evident from the work itself; the truth is far different. For another implication for the dating of one of Augustine's works, see footnote 15 above.