…. siue trans altas gradietur Alpes
Caesaris uisens monimenta magni
Gallicum Rhenum horribilesque ultimosque Britannos.
So the MSS (except that for ultimosque O has the uox nihili ‘uitimosque’ and that a corrector of R and some inferior MSS omit the first -que); but in modern times only the Budé text of G.Lafaye has printed horribilesque unemended and unathetised, and that without a note despite Bentley's observation (on Hor. C. 3.14.11) that there is no instance of a short final vowel in hiatus where there is no break in the sense. The MSS reading does not scan.
Statius in his editio prior of 1566 saved the metre by introducing a postponed et – horribilesque et ultimosque, taking the first -que to mean ‘both’. Scaliger preferred to substitute the postponed et for the first -que, and was followed by Voss and Bentley (on Hor. C. 1.35.29). There is no need to detail here the numerous other suggestions that have been made: I agree with Fordyce that none of them is entirely satisfactory. Some of them substitute a noun for the first -que, and one of these, Haupt's horribile aequor (Opuscula I. 98-105), has now won general favour, having been adopted by Postgate, Merrill, Kroll and Mynors. Kroll admits that horribiles would be a suitable enough epithet for the Britons cf. Caes. B.G. 5.14.2; and for the epithet standing alone, Hor. C. 1.29.4); but the reason given by Haupt for introducing an alien noun is that horribiles detracts from the force of ultimos (Fordyce speaks vaguely of ‘the awkwardly paired epithets’).