130 
On the Letter *r.' 
latter of these has lost a radical r is shown by the Greek equivalents 
Tvppr]voif Tvpffrjvoi, and yet more distinctly by the variety — and ap- 
parently the oldest form of the name — Turscus, as it is found in the 
Eugubine Tables. And Bomilcar, we are told, is equivalent to Bar- 
milcar = A.S. ^theling. 
Lastly, in the Platt-deutsch and in Old Saxon we find we for wer 
and de for der, both as an article and a relative ; as in Reineke de 
Vos, 
De koniug sprak : we is so sot 
De Reinken dor bringeu dat dridde bodl 
So akin to this de we have in A.S. both se and ])e, and our he is the 
her of - Tat. and er of modern German ; but all these are merely in- 
stances of that abrasion which all terminations in the Lido- Germanic 
languages are constantly undergoing. 
One characteristic of this consonant which has been frequently 
noticed is its tendency to change places with its attendant vowel. 
In the Devonshire dialect there are many instances of this, as urn 
for run, urd for red, urch for rich, lurches for breeches, curst for 
crust, Kursmis for Christmas, irt for right, girt for great, apern for 
apron, urdgment for regiment, firnt for front, purceed, pursarve, pur- 
vent, purmote, purtect, for proceed, preserve, &c. So I suspect that 
the noun rod, in its various senses, with its by-form rood, also used 
in a variety of meanings, and the German ruthe, are all akin to the 
A.S. gyrd, geard, &c. — rod or staff. Germ, gerte, Dut. gaard, Friesic 
gerd, early English yerd, modern English yard (this being probably 
quite distinct from yard, as akin to garden, hortus, court, &c.) ; and 
that in yard, as compared with rod, we have another instance of the 
metathesis now under consideration. The loss of the semivowel y 
before the r is to be exjDlained by the simple impossibility of pro- 
nouncing it, unless indeed it were vocalized into i ; but in all the cases 
we are now considering, the change of pronunciation involves no 
change in the number of syllables. Or if the g of gyrd was pro- 
nounced as we now sound it in the verb to gird, the loss of the mute 
in rod is but analogous to the disappearance of the c in repo, as 
compared with the English creep. 
The fact of this metathesis is not incapable of explanation, if at 
least we suppose that it is in every case the vibrant r that we are 
dealing with (and we have no evidence that a thousand or even five 
hundred years ago the non-vibrant r existed in our language) ; for if 
we pronounce the rough vibrant r in yrnan, cirspan, or crisp, gars or 
grass, &c., and pronounce it strongly, we at once find that this r will 
not admit of any other consonant so closely adhering to it, either 
before or after it, that there shall not be a brief vowel sound inter- 
vening. Sometimes (as in run, i-ron) this is the vowel which has 
been above shown to be nearest akin to the r. More commonly it is 
the vowel of the root which seems to prolong itself beyond the vibra- 
