Two words in the centre of Caesar's evocation of Massilia on the brink of disaster have spurred numerous emendations: publicis custodiisque. Meusel lists roughly twenty repairs, most of which I disregard because they contravene one of the few certainties here, namely, the fact that the aut muro . . . aut templa deorum construction makes it unlikely that what precedes muro is a place.
The simplest repair is Carter's inversion of the two words. This improves upon an idea already present in the Aldine edition (publicisque custodiis). Inversion is a more plausible form of corruption than the migration of -que from publicis to custodiis, and we saw earlier that inversions are to be expected in our archetype. But the rhetoric is poor: custodiis ('guards') should not conclude a list headed by the aged, the children, and the women. Besides, the city's garrison may have been mentioned earlier in the sentence, in omnis iuuentus quae in oppido remanserat (more on this below). Furthermore, the expression custodiae publicae gives pause. There is nothing comparable in Caesar, who uses custodiae more than a dozen times; the closest parallel is a reference to public slaves in Pliny (Ep. 10.19.1 per publicos ciuitatium seruos . . . asseruare custodias), but that passage does not illuminate this one. Elsewhere only custodia, in the singular, is modified by publica, and then it means 'custody' (TLL 4.1560.1–14).
Nipperdey gave the passage a thorough overhaul, moving custodiis into the relative clause to explain the presence of the iuuentus, excising -que, converting publicis to supplicis (sc. manus), and placing aut before supplicis. The result makes good sense (for the suppliant pg 182hands cf. BC 2.11.4 ad legatos atque exercitum supplices manus tendunt), but the sequence of corruptions that it implies isn't a bit plausible. Hoffmann and others follow his lead but move both publicis and custodiis into the relative clause.12 This is a somewhat cleaner repair, but we've already seen that publicis custodiis is not likely to be a Caesarian expression. Paul, by contrast, develops the idea of supplication by emending publicis custodiisqu(a)e into an expression modifying the city's old men, supplices sordidatique (or consternatique). This requires fewer moving parts but even so is rather implausible.
Hoffmann supposes that publicis custodiis was omitted from the text at some point, then supplied in the margin, then supplemented by -que and inserted into the text at the wrong location. It seems to me that Dübner's simple excision is a better repair. Caesar's rather vague description of the iuuentus here might well have prompted a gloss explaining its function,13 and the peculiar expression publicae custodiae can be more easily foisted upon a glossator than on Caesar himself. The connective will have been added when the gloss was moved into the text.