Peter Hearth makes reference to the North-West Block theory (notably heralded by Maurits Gysseling and Hans Kuhn) which holds that ancient populations living in modern Benelux and northern France were speaking an Indo-European language neither Celtic or Germanic but intermediate geographically and maybe linguistically between these.

This NWB would be evidenced by a set of linguistic particularities in toponymics (and especially hydronimics, considered as stabler and older), notably the permanence of /p/ (that either gave /f/ in Germanic or disappeared in Celtic but present in ancient linguistic elements in the region, but also other forms a priori not expected in mainland Celtic such as the permanence of /k/ or /kw/ instead of its evolution into /p/.

It could either be understood as a Bronze Age or Early Iron Age substrate for later Germanic or Celtic particularities in the region, or even as evidence for a regional identity merely submersed first by a Celtic or Germanic-speaking elite, and later Romanization, and a basis for the existence of a distinct Belgian language.

That said, it's worth pointing out this is not a mainstream theory and that most specialists rather assume ancient linguistic regional elements are virtually wholly understandable as mainland Celtic and can be found or hypothesised outside the area (in Gaul, British Isles, etc.) with few elements of firm differentiation (exception made, notably, of the permanence of a /-mn-/) with, likewise, ancient Belgian identity itself being explainable comparatively to the rest of La Tenian horizon. The broad agreement would rather still be on a mostly Celtic-speaking region in northern Gaul but also in southern Germania at the turn of the common era.

It is still an healthy reminder that, first, the neat division between Germani and Galli set by Caesar along the Rhine is at best a convenient simplification and that the use of these monikers by the general shouldn't imply a linguistic or ethnic affiliation.

But it also points that our knowledge of pre-Roman languages is extremely fragmentary : at the exception of late epigraphic evidence for Gaulish, we are badly informed on transalpine languages and dependent from Roman-era toponymic lists. We know that non-Celtic languages were spoken in ancient Gaul, namely in Aquitaine, and nothing prevents otherwise unknown languages (in Belgica, but also in the Alps) to have existed alongside or as substrates in itself and explaining particularities that do not fit broad hypothesised evolution or cannot easily be found in other regions.

The context, for those not familiar, is as follows (this is p.53):

While the territory of ancient Germania was clearly dominated in a political sense by Germanic-speaking groups, it has emerged that the population of this vast territory was far from entirely Germanic. In the great era of nationalism, anywhere that threw up plausibly ancient Germanic remains was claimed as part of an ancient and greater German homeland. Analysis of river names has shown, however, that there was once in northern Europe a third population group with its own Indo-European language, located between the Celts and the Germani. These people were under the domination of the other two long before Roman commentators reached the area, and we know nothing about them.

There are no footnotes, of course, so I can't be sure what exactly he is referencing here. I believe he might be talking about "Old European hydronymy," championed by Hans Krahe in the 1960s. The theory seeks a Mesolithic (or early Neolithic?) language across Europe, from Britain to Iberia to the Baltic, which can account for an apparent similarity in the names of rivers throughout Europe. Mallory and Adams (2006, 130) sum it up neatly as follows:

As for the rivers, there is a vast literature on the river names of Europe and Asia that has attempted to discern both a system of river names and, often, their origin. Much of modern discussion takes Hans Krahe’s ‘Alteuropaisch’ as its point of departure. Krahe envisaged a hydronymic system that embraced the linguistic ancestor of what we might term the North-West Indo-European languages coupled with Messapic and Venetic. This system was extended back to Proto-Indo-European by W. P. Schmid, while more recently much of the same hydronymic system has been ascribed to Basque by Theo Venneman. All these systems are comprised of a wide variety of river names that are generally derived from exceedingly small bases (conjectural roots such as *el-, *al-, *er-, *or-, etc.) that may belong to any number of different languages or language families and whose underlying meaning simply cannot be verified to any confident degree. The actual number of river names that can be reasonably reconstructed to Proto-Indo-European, as we have seen above, is extremely few.

If this is what he is talking about here, and I am guessing it must be. There is no other secretive IE branch lurking in the north of Europe, and if there were, Mallory and Adams would at least mention it. They mention other possible or probable IE candidates for which there is just not enough good evidence, like Lusitanian, Illyrian, Venetic, North Picene, Rhaetic, Phrygian, Thracian, etc. There is certainly no reason why a lost IE language could not exist between Celtic and Germanic, but I would wonder what the evidence is.