Yan, Tan, Tethera...

The counting methods of North England shepherds are ancient relics from the time when all of England was Celtic in speech. The language of the land was Common Brittonic, the parent language of Welsh and Cumbric (among other descendents). Eventually, Common Brittonic was replaced in most of England by Anglo-Saxon (or its descendant Middle English). Today we're gonna have some fun and give the phonological developments of these preserved counting methods.

Common Brittonic1.
"one"> *oinos (m.), *oinā (f.)> *ainā> *ain*jain> *jān> *jan
Rathmell, Eskdale aen ; Wilts ain ; Lakes aunBowland, Nidderdale, Derbyshire, Dales yain ; Ayrshire yintyWensleydale yahnKeswick, Westmorland, yan
"two"*dewou (m.), *dewī (f.)> *tewī> *tej-(?)> *t(e)(j)- + number one by analogy
All dialects
"three"*trīs (m.), tiserīs (f.)> *titherī-(s?)> *tetherī> *tethera> *teCeCV (C and V matched to C and V in 'four' by analogy)*t > *0
Wensleydale titherAyrshire thetheri ; Westmorland tetherieAll other dialectsMost dialectsBowland eddera ; Nidderdale, Derbyshire eddero

By this 30 minute reconstruction, we can probably guess that the most-recent common ancestors to shepherd "one," "two," and "three" were *ain, *tej-?, and **titherī-?. If we then compare the numbers between each other we can also discover some interesting things.

First is that counting methods utilized the feminine form of the number, unusual in that the masculine form is usually conserved among IE languages. This should actually be unsurprising as the word for sheep was feminine (Common Brittonic (f.) *dametas "sheep" (lit. "little ox") > Old Cornish f. dauat  ; Welsh f. defad ). Thus, a post-Common Brittonic forms of 1,2,3 were *ainā, *tewī, and *tetherī. 

Second is that devoicing of *d > *t must have been a very early change as it is found without exception across the different counts (and in ancient Cumbric words). We would not expect such universal conformity if the change were a late-date devoicing due to influence from English two.

Third is that analogically modifying the numbers (for example, changing the word for "two" to more closely mimick "one") must have happened at a relatively late date and it was certainly not a universal process.