Page xix

A NOTE 
ON TRANSLITERATION 
AND NOMENCLATURE 
 One of the obvious problems to be solved by the editors of such a work as
this, intended both for general readers and for scholars in many different
disciplines, is how to render the names of persons and places, and a few
other terms, originating in languages and scripts unfamiliar to the English-speaking
reader and, indeed, to most readers whose native languages are European.
In the present volume, and presumably in the entire work, these comprise
principally Arabic, Turkish, Persian, and Armenian, none of which was normally
written in our Latin alphabet until its adoption by Turkey in 1928. The analogous
problem of Byzantine Greek names and terms has been handled by using the
familiar Latin equivalents, Anglicized Greek, or, occasionally, Greek type,
as has seemed appropriate in each instance, but a broader approach is desirable
for the other languages under consideration. 
 The somewhat contradictory criteria applied are ease of recognition and
readability on the one hand, and scientific accuracy and consistency on the
other. It has proved possible to reconcile these, and to standardize the
great variety of forms in which identical names have been submitted to us
by different contributors, through constant consultation with specialists
in each language, research in the sources, and adherence to systems conforming
to the requirements of each language. 
 Of these Arabic presents the fewest difficulties, since the script in which
it is written is admirably suited to the classical language. The basic system
used, with minor variants, by all English-speaking scholars was restudied
and found entirely satisfactory, with the slight modifications noted. The
chief alternative system, in which every Arabic consonant is represented
by a single Latin character (t for th, h for kh, d for dh, s for sh, g for
gh :(<see image>)) was rejected for several reasons, needless proliferation
of diacritical 
xix