132 
On the Letter ' r.' 
This too is easy to account for. The only difference in the formation 
of these sounds is, that in the r the breath passes continuously, or 
with numerous momentary interruptions, over the tip of the tongue, 
and in the I it passes by the tongue on each side, the tip touching 
the palate. 
This diiference, though slight in itself, we can with perfect readi- 
ness both make with the tongue, and apprehend with the ear. But 
it is not so with all the members of the human family. , Some of the 
South Sea islanders for instance are unable to distinguish these 
letters either in speaking or by the ear. A lady of my acquaintance, 
who lived for several years at Tahiti, had two little girls one named 
Mary and the other Emily. These names were a source of great 
embarrassment to the natives, who in pronouncing them made them 
both Amala ; with which we may compare the rpavXiffjuoQ ascribed 
to Alcibiades by Aristophanes (VespaB, 45), oXcig for opag, and so on. 
We may also notice that in the early Welsh poetry a final I will very 
often rhyme with a final r ; while in some parts of Spain, especially 
in Andalucia, these two liquids are pronounced so much alike as to 
be not unfrequently confounded by uneducated persons ; thus they 
will write terrol or tellor for terror, hollol for horror. 
The actual change of one of these letters into the other takes place 
frequently in immediate juxtaposition with that other ; as the Sans- 
krit tdrd shows that the r is radical in the English star, Gr. aarrjp, 
&c. ; but when to this root the Latin diminutive termination -la is 
added, it is not ster-la or ster-ula that results, but Stella. Sometimes 
the / is changed into r on account of the vicinity of another I, as in 
familiaris for familialis : but the presence of another / is not essen- 
tial. We have no diminutive termination -ly in English, and Molly, 
Dolly, Sally, are formed by a direct change of the consonant from 
Mary, Dora, Sarah. 
The Hebrew nnu^nu^ (shar'sh'rah), a chain, Chaldee rhmhm 
(shal'sh'lah) ; mDon« (ar'man5th), palaces, later Hebrew m3o'?« 
(al'manoth), — and many more examples might be cited from the Se- 
mitic languages ; the Grk. apyog, Lat. albus; ^^ayiWiov as the New 
Test, form of the Lat. jiagellum ; the coexistent Latin forms celeber 
and creber, Unealis and linearis; the Romance citola from Lat. cithara ; 
alblast, an Old English corruption of arhalast, a crossbow {arcubalis- 
ta) ; palfrey, Fr. palefroi, in medieval Latin parafredus ; the Wala- 
chian poporu from the Latin populus, people ; Salisbury as the modern 
form of Sarisbury ; — all of these exhibit the same direct interchange 
of r and /. 
The afiinity of r and I being allowed as a well established fact, and 
being also easily accounted for physiologically, it is not difficult to 
see that r and d will also interchange. The latter consonant difiers 
from I only in the lateral expansion of the tongue, to intercept the 
breath, which in the I is allowed exit on each side of the tongue. 
