218 MR W. F. SKENE ON THE CELTIC TOPOGRAPHY OF SCOTLAND, 



where we have Loch Leven with the two Lomonds on the side of it, and the river 

 Leven flowing from it through Strathleven. This recurrence of the same 

 words in connection would be unaccountable were it not an example of the same 

 thing. Leven comes from the Gaelic Leamhan, signifying an elm tree, but the 

 old form is Leoman, and the m becomes aspirated in a later stage of the language 

 and forms Leamhan, pronounced Leven. Here the old form adheres to the 

 mountain, while the river adopts the more modern. 



A curious illustration of two different terms lying side by side, which are de- 

 rived from the same word undergoing different changes, will be found in Forfar- 

 shire, where the term Llan for a church appears, as in Lantrethin. It is a 

 phonetic law between Latin and Celtic, that words beginning in the former with 

 pi are in the latter //. The word Planum, in Latin signifying any culti- 

 vated spot, in contradistinction from a desert spot, and which, according to Du- 

 cange, came to signify Cimiterium, becomes in Celtic Llan, the old meaning of 

 which was a fertile spot, as well as a church. In the inquisition, in the reign of 

 David I., into the possessions of the See of Glasgow, we find the word in its oldest 

 form in the name Planmichael, now Carmichael ; and as we find Ballin corrupted 

 into Ban, as Ballindoch becomes Bandoch, so Plan becomes corrupted into Pan, 

 and we find it in this form likewise, in Forfarshire, in Panmure and Panbride. In 

 the Lothians and the Merse this word has become Long, as in Longnewton and 

 Longniddrie. 



The Celtic topography of Scotland thus resembles a palimpsest, in which an 

 older form is found behind the more modern writing. I shall not detain the 

 Society further by going through other examples. The existence of the pheno- 

 menon is sufficiently indicated by those I have brought forward, and I shall 

 conclude by stating shortly the results of this investigation. 



1st, In order to draw a correct inference from the names of places as to the 

 ethnological character of the people who imposed them, it is necessary to obtain 

 the old form of the name before it became corrupted, and to analyse it according 

 to the philological laws of the language to which it belongs. 



2d, A comparison of the generic terms affords the best test for discriminating 

 between the different dialects to which they belong, and for this comparison it is 

 necessary to have a correct table of their geographical distribution. 



3d, Difference between the generic terms in different parts of the country 

 may arise from their belonging to a different stage of the same language, or from 

 a capricious selection of different synonyms by different tribes. 



4th, In order to afford a test for discriminating between dialects, the generic 

 terms must contain within them those sounds which are differently affected by 

 the phonetic laws of each dialect. 



oth, Applying this test, the generic terms do not show the existence of a 

 Kymric language north of the Forth. 



