Examples of breaking the circular reasoning  :
Vindolanda , Lincoln

(1) all of Britain was Celtic
(2) this implies that all (Roman) place-names in Britain are of Celtic origin
(2) therefore, only etymologies in Celtic will be accepted
(3) all these etymologies together are the very proof of (1)
(4) which confirms also statement (2)

The CBA is a classic example of circular reasoning.  We try to break the circle by demonstrating that most Celtic etymologies are bogus and that 'tertium' (a third party) is possible: a Germanic etymology which gives a much sounder etymology.

Here are some examples of how it can be completely different. Both etymologies of Vindolanda and Lincoln are officially in proto-Welsh. This section illustrates that there is an alternative.

Vindolanda is a Roman fort near Hadrian's Wall. The official etymology of the place-name is in Welsh: vindo+lann. 'vindo' = white +'llan' = land, fields.

The first problem lays within the word 'llan'. This is clearly the same word as 'land'. 'Land' is to be found in all German languages: middle Dutch, old high German, old Saxon, old Fries, old English, Gothic. Outside the German languages: old Irish: 'land', Welsh: 'llan', old Prussic: 'lindan' (valley), old Russian: 'ljadina' (shrubs, weed). The geographically limited spread of the word points towards a German substrate word [1], and not a PIE word. The word must have been introduced in Old Welsh first (where the 'd' disappeared), and later in Old Irish (where the 'd' was maintained). 'Land' does not occur outside the Germanic zone on the Continent. It is not a Gallic (=Celtic!) word. The Old Prussian and Russian versions are known to be loanwords.

In all Brythonic dialects, a llan, land or lann is a loanword and always means land owned by the church or some Catholic propriety.  So, it is obvious that the word was introduced together with the Christian religion. Clearly, it is not an original Celtic word.

The problem in Vindolanda is that the word 'landa' is too early. When Hadrian's Wall was build the Catholic Church was still some sort of expanded former-fishermen's-club. The Romans invented nothing here, they simply took over a local name. And who knows how old that place-name was at that moment.

The second problem is the word 'vindo-'  = 'white' and what it refers to (Modern Welsh: gwyn-, also in the first name Gwendolyn = 'white circle' = the moon). The Brythonic meaning of Vindolanda, 'white fields', is at least doubtful. Many objects in leather were discovered on the site because the soil is black due to a lack of oxygen.

Vindolanda is Germanic

The etymology in proto-English is far more probable:
(1) Vindo means slightly raised in old Norse. The word is related to the verb 'to wind' in the sense of 'to turn, curve up'. It was pronounced 'winde-'.
(2) Land is typical Germanic and means a good and open place to build a home (farm) upon, the farm building itself. [2]  Here it means 'farm building'.

Vindolanda is a compound word, so it has one single and specific meaning.  This too excludes the 'whitefield' meaning. Why would a farmhouse be 'turned up' on poles or raised? The answer is simple: it was a common granary. The only way to preserve grain for a longer time is to store it high and dry, above the ground on a plank floor. A granary was a more expensive building as it had a full plank floor above the ground. As storage places, granaries were important.  Farmers kept a stock of grain for themselves in small granaries, so the larger granaries stored the local surplus. Merchants need not to visit each farm in search of a merchantable surplus, all they had to visit was the local common granary. [3] No wonder that those places often became villages or cities.

Vindolanda means 'raised granary'

Vindolanda is (proto-English) Brigantes territory

Vindolanda is situated in the middle of Hadrian's Wall, circa 35 miles from Newcastle-upon-Tyne (east coast in the north of England) and circa 35 miles of Carlisle (north-west coast). Its proto-English origin situates it or near the old language border, or can be an indication that proto-English had already moved up westwards before the Roman era. Ptolemy (second century AD) names nine towns as belonging to the north-British Brigantes: mainly in Yorkshire, but also north-Yorkshire, Northumberland AND in Cumbria (west of the Pennines). The later kingdom of Northumbria would control the same region. Its capital was York. Simple coincidence? It's more likely that in this region, the east dominated the west since pre-Roman times and continued to do so afterwards.

The Brythonic name Brigantes is linguistically identical to the Germanic name Burgundes. It is likely that the Romans had Gallic (Brythonic) speaking interpreters. Therefore, the Brythonic version of the local names was the one that was written down. The Germanic word  Burgundes meant originally 'people from the fortified villages on the hills',  from berg = hill, and was later expanded as 'citizens, burgh or bury people, villagers'.  The Brigant capital was York (Civitas Eburacum). 

In the Germanic world, nearly all villages were 'burghs', even the ones build on flat plains. The word had simply evolved to become a word for town, village or city (the latter two words came later and are of French origin). In Dutch and German berg means hill and -burg means town. In burg the connotation on a hill had faded away. The Brythonic word is suspicious because the Brythonic speakers had not the tradition to name their villages 'on a hill' (= brig).  Villages or cities in Welsh are called caer not brig. The word Brigantes means hill people and not villagers. Brigantes would literally refer to the dwellers in the Pennine hills, the main low mountain ridge in the region.  Yet the regions with real power were always situated in the more fertile valleys.  Nearly all villages at the time were build in valleys or on the plains. Fact is that the region around York is a wide and flat valley. Fact is that the region around York was the real power house of the so-called Brigantes. Therefore, it is very likely that the word Brigantes is a literal Brythonic translation of the Germanic Burgundes (= 'our citizens') and that it had no 'hillbilly' meaning. Which means that the Germanic word is the original one. Sadly, the Romans stuck to Brigantes.

Lincoln is a Germanic place-name too

Lincoln. The Roman name for Lincoln is well attested in the classical geographies. The name in Latin was Lindum Colonia or Lindensium Colonia (both are attested). "-coln"= colonia = a resting place for legionnaires.
In Welsh (or Brythonic) Lindum is generally supposed to mean “dark water” or Blackpool. The capital of Ireland Dublin means exactly the same: Dubh = dark, black + lin = pool. Here noun and adjective are reversed, which on itself is strange. Welsh: llyn=‘lake’ and Gaelic: linne=’pool’. '–dum' gives more difficulties. It could be '–dun'= ‘fort’, ‘castle’ or '–du'= ‘black’, ‘dark’. "-dun" is rejected because the Roman version would then be "-dunum". So, Lindum = Poolblack.
This is acceptable for Lindum, but its synonym Lindensium has a problem. An 'e' and 'i' are often interchanged. Those vowels are close when they occur in foreign (from the Roman point of view) languages. Confusing an 'e' with an 'u' is less likely. Lindensium cannot been explained in Brythonic.

If one supposes that the name was of proto-English origin, then all is very simple: Linden or Lime (tree) is the meaning. Lindens can grow for hundreds of years. This counts as a landmark. The original word is linde (singular; it’s the same word in Dutch or German and a substrate word limited to the German languages). Linden is a (forgotten in modern English) plural. Lindensium could be Latinised short for ‘next to where linden grow’. 

This is possible because:
(1) Many English place-names refer to trees (e.g. Sevenoaks).
(2) Linden-sium is explained.
(3) Lindum means: linde (singular) + um (for declension).

The Roman fort was centered on a 60 meter high hilltop at the end of a limestone ridge and allegedly overlooked a hypothetical (and dark) pool in the river Witham from the north.
So, the trees upon the slope were limes (trees) or linden. The hilltop itself was not occupied by civilians. Hilltops can be very cold and windy and not well suited for living upon. The hamlet (Linden) must have existed on the sheltered slope facing south before the Roman camp was build, above the hamlet, and overlooking it.
I use Lincoln to demonstrate that a proto-English etymology is very well possible.

Map of Lincoln.

If Lincoln can be explained using proto-Welsh, the etymology of Witham (its river) remains officially a mystery. The old name was Wimme, Wemme (AD 1086). A possible explanation is a relation with old Saxon wemmian = 'bubble up', old High German wimmen = 'to move', modern Dutch wemelen = 'many movements, to swarm'.

So: Witham=moving waters. This explanation is straightforward and must have been discovered for long, but was rejected because an explanation in Welsh was mandatory.

Combine Witham and Lincoln, and it's clear that the region was proto-English before the Romans settled.