598 
Proceedings of the Royal Society 
It is remarkable that in the later Romance languages, Z, when it 
is found in Latin, sometimes disappears, and is replaced by r: as 
apostolus, apotre; epistola, epitre ; capitulum, chapitre, &c. In- 
deed l does not very well stand its ground in modern times. In 
Italian it often becomes an i ; in French it becomes an u ; and in 
the lower German dialects, such as our own Scottish, it is similarly 
changed or lost. 
Let us now assume as an interim hypothesis that r and l are 
interchangeable in Greek and Latin, and see if that assumption 
will afford us results that tend to confirm its truth. 
The names for the swallow in those two languages are respec- 
tively and hirundo. Upon the hypothesis suggested, 
XsXlSuv may be changed into xepiSwv; and then, by well-known 
tendencies of the Latin language, the final v will be dropped, 
leaving xqoiSw, while an n may be inserted before the d to 
strengthen the syllable, as in tundo, tudi ; fundo, fudi ; fmdo, fidi ; 
frago, frawgo ; tago, tango, &c. We thus get ^epn/Sto and hirundo, 
the identity of which is manifest. 
XaAa£a and grando, the words, for hail, may be assimilated 
nearly in the same way. XaAa£a becomes x a P a t a : this when con- 
tracted becomes xpa£a, as X a P L * becomes gratia. Z is = to ds or 
di, and with an inserted n, x a P a £ a is equal to grandia, which is 
close upon grando. 
Upon this footing we see the identity or near affinity of Kpvimo 
and KaAu7TT<o; and with these, perhaps, kXcttto) may be connected. 
KvkAos, the Greek for a wheel or ring, may in its more primitive 
form be set down as KVKpos, which seems cognate to the Indian 
form chakra, with the same meaning. But Ka/cpos with a slight 
metathesis leads easily to the Latin circus, circulus ; and it is 
again possible that by aspirating and modifying the consonants, 
circus becomes identified with the Teutonic ring <= hring, while 
kvkXos is thought to be cognate to the Teutonic wheel ; so many 
diversities of form may thus be derived from the same elements of 
a guttural twice repeated, and a liquid r or l variously arranged. 
f EAj uivPs, by changing the l into r and prefixing a digamma, 
becomes vermis, the relation of the aspirate and digamma being 
the same as in icnrepos and vesper. The Greek iSpa would be 
easily changed in Latin into sedla , which by assimilation becomes 
