The Structure of Tacitus' Annals

Three Hexads or Two 'Ogdoads'?

C.A.M.W.S. (Nashville), April 11, 1996

© M. Hendry 1996-1999

Like the larger works of all the other major Roman historians, the Annals and Histories of Tacitus survive only in part. We know (from St. Jerome) that they totalled 30 books, but we do not know how these were distributed between the two works, although it seems likely that there were either 18 books of Annals and 12 of Histories or 16 of Annals and 14 of Histories.

Figure 1: Hexads

Two opinions are current on the structure of the Annals:

In chapter xxi and Appendix 35 of Tacitus, Ronald Syme argued that there were 18 books, divided into 'hexads' or groups of six, with the six books on Tiberius forming the first hexad, two on Gaius plus four on Claudius forming the second, and six on Nero forming the third. Syme's analysis is depicted in the first illustration: extant parts are shaded. That would leave 12 books for the Histories, and these might also be divided into hexads, with one for Galba, Otho, Vitellius, and Vespasian, the other for Titus and Domitian. Besides overall neatness, one advantage of Syme's scheme is that the last two years of Nero's reign, after our text breaks off in the middle of Book 16, were full of exciting and important events. It would have been difficult to squeeze them all into half a book. All the more so, if the Annals continued on past Nero's death to cover the second half of the year 68, so as to link up with the beginning of the Histories. (On the other hand, even if it did, two and a half books seems excessive.) One obvious disadvantage of the hexad theory, which Syme rather glosses over, is that Gaius and Claudius make quite a disparate pair. In general, Syme seems much less interested in his second hexad than the first and third, no doubt partly because not much of it survives.

The second opinion on the structure is less positive, and needs no diagram to illustrate it. Against Syme, others, such as F. R. D. Goodyear in his unfinished commentary on the Tiberius books of the Annals, argue that there were 16 books, not 18, and do not worry much about any grand structure, or at least any structure larger than the division by emperors. One advantage to this position is that our only actual evidence supports a total of 16 books for the Annals. Some manuscripts number the surviving books of the Histories as if they were books 17 through 21 of the Annals. If we then go back to Syme's alternative hypothesis, that Tacitus intended to write 18 books of Annals, but only lived to finish 16, that still leaves 14 books for the Histories, which means that we cannot have hexads in both works.

Figure 2: Ogdoads

I would like to propose a third possibility, which combines the advantages of the other two. I suggest that Tacitus wrote 16 books of Annals (thus Goodyear), with a neat and clear structure of blocks of books (thus Syme), consisting not of three 'hexads' but of two 'ogdoads' or groups of eight. (As awkward as it sounds in English, the word 'ogdoad' has a good Greek pedigree: it is of course Neoplatonic, like 'ennead'.) The first ogdoad would consist of six books for Tiberius plus two for Gaius, the second of four each for Claudius and Nero. My analysis is depicted in the second chart.

The principal difficulty in proving — or for that matter disproving — my analysis of the structure of the Annals is that there are so few corresponding parts available. What I mean is this. If, as Syme argued, Tacitus wrote three hexads, then we have the beginning and end of the first, and most of what comes in between, the end of the second, and the beginning of the third. That is a good deal to work with. If, as I think, he wrote two ogdoads, we have the beginning but not the end of the first, and neither beginning nor end of the second, though we do have more than half of what comes in between.

My basic hypothesis is simple: that the four emperors covered in Tacitus' Annals are best taken in two pairs, in interlocking order, with Tiberius corresponding to Claudius and Gaius to Nero. The similarities between Tiberius and Claudius are quite striking, as are those between Gaius and Nero.

To begin with, Tiberius and Claudius were old men, Gaius and Nero very young: over-fifties and under-thirties, we might say. Here are the significant dates and ages (in bold):

Birth: 42 B.C.E. November 16 Birth: 10 B.C.E. August 1
Accession: 14 C.E. September 17 55 Accession: 41 C.E. January 26 49
Death: 37 C.E. March 16 77 Death: 54 C.E. October 13 63
Birth: 12 C.E. August 31 Birth: 37 C.E. December 16
Accession: 37 C.E. March 16 24 Accession: 54 C.E. October 13 16
Death: 41 C.E. January 24 28 Death: 68 C.E. June 9 30

In each case, the older man was succeeded by his great-nephew, who was also his adopted son. Each of the older pair became emperor by default, because no one else was available, Tiberius adopted after all of Augustus' other heirs had died, Claudius hauled out from behind a curtain by the Praetorians after the murder of Gaius. Each of the older pair had spent years disregarded and despised, when it was assumed that he would never inherit, Tiberius sulking on Rhodes, Claudius drooling in his study. Each of the younger pair came to power more or less legitimately, according to his own expectations and the plans of his predecessor — as of course Tiberius did, in the long run. Each was paired with a younger and more directly legitimate heir, Tiberius Gemellus and Britannicus, whom he very soon disposed of. Tiberius and Claudius were both murdered (if they were murdered) in domestic plots, by smothering or poisoning. Their successors were overthrown openly, Gaius stabbed to death in a coup d'état, Nero committing suicide after whole provinces had revolted against him and inaugurated a civil war that continued for a year after his death.

Tiberius and Claudius shared military skills or (in the latter case) military interests, which Gaius and Nero despised. We might contrast Tiberius' many battles (before his accession) and Claudius' (vicarious) conquest of Britain with Gaius' seashell-gathering expedition. Tiberius and Claudius also had pretensions to republicanism. Each made a show of his supposed reluctance to rule and eagerness to share his burden with the Senate. Gaius and Nero were open practicioners of oriental despotism, demanding to be worshipped as gods, a demand which in each case led to a Jewish revolt. Tiberius and Claudius had pedantic and old-fashioned literary tastes, while Nero at least (if not Gaius) was thoroughly avant-garde.

Both Gaius and Nero were accused of incest, Gaius with all three of his sisters, Nero with his mother. Tiberius spent much of his reign 'out of town', while Claudius was 'out to lunch' during his. Each of this older pair allowed someone else (Sejanus and Agrippina) to run the empire while he was off on Capreae or in a stupor, as the case may be, and in each case the trusted lieutenant tried to take over much more power than had been granted, and was murdered for it, with the help of more trustworthy henchmen. By contrast, the governments of Gaius and Nero were far too 'hands-on' for most people's liking — or safety. (Agrippina, of course, was not disposed of until she had succeeded in changing the succession, and then overreached herself in dealing with her son. That is the main reason Claudius only gets four books as against Tiberius' six.)

Tiberius and Claudius were frugal, Gaius and Nero extravagant. Each of the latter left the empire leaderless as well as bankrupt, with two days of anarchy (no emperor at all) following the assassination of Gaius, and a full year of bloody chaos (no fewer than four emperors) following the death of Nero. In my view, the (missing) transition between books 8 and 9, after the assassination of Gaius, when the Republic had been in effect restored by default, would have been the high point and the central turning-point of the Annals, as the Senate, handed power on a platter, showed itself totally incapable of ruling Rome or even selecting a decent emperor, or any emperor at all — they dithered until the decision was taken out of their hands. As Tacitus describes a similar situation in Armenia (Ann. 2.4), Rome was then magis sine domino quam in libertate, 'more without a master than in a state of freedom'.

At this point, having taken up most of my allotted time in showing that Tacitus should have written his Annals in this way, we need to ask whether he did so, and that is a much more difficult question to answer. To prove my case will take a substantial monograph. Three small hints may suffice here:

First, as Goodyear notes, Tacitus appears to group the emperors in pairs when he first names them (Ann. 1.1.2): Tiberii Gaique et Claudii ac Neronis.

Second, Syme divides his hexads into three-book halves, though it only really works for the first, where Sejanus is ostentatiously introduced at the beginning of Book 4. However, Tacitus' obituary of Tiberius (6.51.3) does not divide his reign into 3 + 3 books, but very nearly into 3 + 1 + 1 + 1:

morum quoque tempora illi diuersa: egregium uita famaque quoad priuatus uel in imperiis sub Augusto fuit; occultum ac subdolum fingendis uirtutibus donec Germanicus ac Drusus superfuere; idem inter bona malaque mixtus incolumi matre; intestabilis saeuitia sed obtectis libidinibus dum Seianum dilexit timuitue: postremo in scelera simul ac dedecora prorupit postquam remoto pudore et metu suo tantum ingenio utebatur.

Since Germanicus dies at the end of Book 2 and Drusus very early in Book 4, their restraining influence is pretty much restricted to the first three books. Tiberius' mother dies at the beginning of Book 5, and Sejanus very probably either at the end of 5 or the beginning of 6, though the passage does not survive. If we can lump together Book 6 (Tiberius' open depravity and cruelty) with the lost Books 7 and 8 (Gaius's similar habits), this will make a scheme of 3 + 1 + 1 + 3 books, which closely matches the second ogdoad, where there are 3 books for Claudius alone, 1 for Claudius under the thumb of Agrippina, 1 for Nero still tied to his mother's apron-strings, and 3 for Nero alone after he kills her — assuming, of course, a total of 16 books. In other words, the middle two books of the first ogdoad (4 and 5) are devoted to Sejanus, of the second (books 12 and 13) to Agrippina, as illustrated on my handout.

Third, as I have mentioned, Syme would like to divide his hexads into halves, but admits that there is no break between books 15 and 16. My theory predicts strong breaks after books 3, 5, 11, and 13, as well as at the ends of reigns, of course, but not at the end of 15, and this suits the evidence much better.

The three small pieces of evidence I have just given are not much to go on. Further arguments will require careful comparison of the extant parts of the Annals to show how Tacitus has shaped his story to fit the scheme I have outlined. There is, of course, a circularity problem here. It is difficult to prove that Tacitus has reshaped or deformed the history of the Julio-Claudians to fit a predetermined pattern when so much of the history of the period is preserved only in his version. However, I wanted to put my hypothesis on record before anyone else thought of it: though it will be hard to prove, it seems so obvious.