Asyndetic pairs dependent on a single preposition have tended to be treated as problematic by editors and scribes, and manuscript variations are not unusual. I have not, however, noted explicit discussions of the issue by editors (see however Preuss 1881: 53). The criteria that may be used by an editor in judging possible cases of asyndeton include structural factors, and an author’s practice elsewhere. I start with a structural pattern and variants, and then refer to two possible asyndeta of this type in Caesar’s Bellum ciuile, which are dealt with in detail in XVIII. Finally, I consider possible examples in several other writers.

By ‘accumulation’ I mean the juxtaposition of one asyndetic pair with another pair, or with more than one other, or the placement of an asyndetic pair alongside or within coordinated groups of various lengths and types. An asyndeton bimembre is in an accumulation if it is not free-standing but is part of a sequence of items that are coordinated syndetically or asyndetically, or have a mixture of both types of coordination. Accumulations have come up frequently already in these introductory chapters, and they will come up more extensively in the later chapters dealing with writers and genres. Accumulations are a prominent location for asyndeta bimembria, such that they are one determinant of this type of asyndeton (see XXXII.3.1). Two words that are usually coordinated may well be used for once asyndetically because they are placed in such a sequence. My aim here is simply to introduce accumulations containing asyndeta bimembria with a little more detail than has been provided so far, but the most comprehensive lists and classifications will be found in later chapters, particularly on Cicero and the historians.

In this chapter I consider asyndeton in Caesar’s Civ., of which there is now a reliable text with apparatus (Damon 2015a). My aim is not merely to classify asyndeta but to consider whether there are criteria that may be used by an editor in deciding on the text when manuscripts have variation between coordination and asyndeton; the chapter is mainly about textual criticism. The passages discussed, which embrace the totality of possible asyndeta (bimembria) in the work, are numbered in bold.

In an earlier chapter (I.3) various ancient commentators were cited on the supposed ‘speed’ or ‘rapidity’ of asyndeta ([Long.] 19.2, Suet. Caes. 37.2, Aquila 41). It is easy to see how such an idea might have come about. An asyndetic sequence as looked at on the written page has fewer words than a syndetic, and fewer words might mean that a shorter time is needed for the articulation.

It is often said, not least in relation to Greek tragedy, that if the two members of an asyndetic pair are not of the same syllabic length the longer term will usually be placed second. The rule for Greek seems to be stated mainly in reference to asyndetic pairs of verbs. Diggle (1994: 99), for example, says: ‘When, in his lyrics, Euripides juxtaposes a pair of verbs in asyndeton, he observes the practice of sound rhetoric: he makes the second verb equal in length to the first, so that it may balance it, or he makes it longer, so that it may outweigh it.’ Twenty-two examples are quoted, with just one case where the rhetorical tendency is not observed. The pairs quite often have the same fore-element. On p. 100 Diggle states that he has included in the list only passages where the verbs are absolute or have a common object.

A compound verb is sometimes placed immediately after its simplex. Such juxtapositions are not all of the same type.

It is sometimes stated or implied that asyndeton is older than syndetic coordination. This is a view that is put bluntly by Timpanaro (1994: 7): ‘Senza dubbio l’asindeto è più antico dell’uso di congiunzioni copulative’. He was aware of the frequency of -que compared with asyndeton in what he calls carmina, and therefore falls back on the idea that the few instances of asyndeton are archaisms of a type that was losing ground (1994: 8). Luiselli (1969: 165–6) likewise says that in carmina syndetic coordination is more common than asyndetic, but suggests that asyndeton was probably more ancient and that it was gradually replaced later by syndetic coordination. But why should asyndeton have been more ancient than explicit coordination? It is to be assumed that underlying the belief is a sense that absence of coordinators must be more ‘primitive’. In fact the coordinator that in Latin has the form -que must be very ancient, as it has cognates in a wide variety of early Indo-European languages (see e.g. De Vaan 2008: 506; also Penney 2005: 40–3, and particularly Dunkel 1982). Dunkel (1982: 141) simply says: ‘Oldest were asyndeton and single *-kwe’.

Pairs of certain semantic types show no more than a weak tendency to occur in asyndeton bimembre: overt coordination is more frequent. There may be a unity of some sort to the terms, which makes them susceptible to occasional asyndetic coordination, but that unity is no more than a background to the asyndeton. It is necessary to look at contexts and structures to see if it is possible to identify factors motivating the absence of a coordinator.

Asyndetic juxtapositions of different tenses or moods of the same verb are very much a feature of legal/official language in Latin. At Cic. Phil. 5.46 (76) the following occurs in a proposal by Cicero for a senatorial decree: libertatem populi Romani defendant defe<nde>rint. By contrast earlier in the same speech two instances of praesideo in different tenses are coordinated, but there the pair is not in a formal motion: 5.37 Galliaque, quae semper praesidet atque praesedit huic imperio, ‘and Gaul, which always protects and has protected this empire’. Even in official language, however, asyndeton is not invariable. In the same speech again (at 5.53) in another such motion we find qui … auctoritatemque huius ordinis defenderint atque defendant. The order of the same two verb forms has been reversed. The coordination gives a better clausula than either (asyndetic) defenderint defendant or defendant defenderint.

In English (for example) when three or more terms are in a coordinated sequence the normal structure is ‘AB and C’, with what is sometimes called ‘end-of-list’ coordination (see e.g. Haspelmath 2004: 572, index s.v. ‘end-of-list’): e.g. bacon, lettuce and tomatoes; Friends, Romans and countrymen. Latin had, up to a point, the same structure (in several forms because of the number of coordinators in the language), but how significant was it, and is it straightforward to interpret? For et and atque used thus (and associated textual problems) see Pinkster (1969). For -que see Adams (2016: 62–4, with some bibliography), with full details for Cato Agr. Pinkster (1969: 266), speaking of the different Latin coordinators found in this structure, says: ‘Que seems to have been most normal in “Classical” Latin, followed by zero, et, atque respectively’. This follows the generalisation: ‘in Latin A B & C is a normal pattern, occurring beside A B C and A B Cque’. According to Dyck (2010: 101) on Cic. S. Rosc.31, citing Pinkster, ‘xyzque appears to be the commonest way of organising three items in Latin’. That is not so. Torrego indeed states (2009: 472) that ‘the marked “A, B, & C” pattern’ ‘is not encountered often in Pre-Classical and Classical Latin, yet it becomes the standard pattern in Romance’, and his examples show that his formulation ‘A, B & C’ was intended to embrace all the Latin coordinators. In fact what Pinkster calls ‘zero’, that is asyndeton throughout (ABCD etc.), is much more common than the use of -que, and also common are what I would call sequences with multiple coordinations, that is with coordinators used within the list, not least to indicate internal unities. I will illustrate this pattern in section 3.

Frequently a coordinated pair (or longer sequence), of adjectives, nouns or other parts of speech, is split up by the insertion of a word or phrase to follow the first member of the coordination (see e.g. Devine and Stephens 2006: 586–91, calling the pattern ‘conjunct hyperbaton’; also Hofmann and Szantyr 1965: 693, Gray 2015: 65–6). A classic example in English is brave man and true. In Latin too it is often homo (or uir) that is the intrusive term, as at Cic. Har. resp. 28: Brogitaro Gallograeco, impuro homini ac nefario, Sest. 56 Brogitaro, impuro homini atque indigno illa religione (in both places with the same referent);1 see further e.g. for homo, Cic. Att. 5.15.3, 5.21.6, Fam. 1.9.19, Prov. cons. 15, etc., and for uir instead of homo, Cic. Fam. 7.18.1 esse fortem uirum et constantem, Fin. 2.80 et bonum uirum et comem et humanum fuisse, Livy 8.8.16 strenuus uir peritusque militiae. At Catull. 12.8–9 (est enim leporum | differtus puer ac facetiarum) puer has the same function.