This is a most valuable suggestion to all indexers. Many persons, to save trouble at the time, write initials instead of full Christian names. It should be a rule always to write these in full. When the index comes to be printed, the Christian names can be contracted if it is necessary to save space. The most important matter in the arrangement of an index is to avoid the confusion of two persons as one, and the possibility of making this blunder is greatly increased by the use of initials instead of full names. In the British Museum Catalogue it has been found necessary in many cases to add particulars to distinguish between men with the same names.

Mr. Curtis goes on to say:

"With regard to the second part of this subject—i.e. the proportions of the several parts of the index—I may observe that the most useful mode of division appears to me to be that which is adopted by many offices—namely, to classify the surname under its first letter, and to subdivide according to the first vowel thereafter, adopting the first subdivision for such names as 'Ash,' 'Epps,' etc., which have no succeeding vowel."

This, however, is a very unnatural arrangement, and has been, I believe, very generally given up. It is therefore unnecessary to refer further to Mr. Curtis's calculations of the proportions of the vowels in the subdivisions. Calculations can be made for the subdivision of the complete alphabet with a better result. Of course, in the case of initial vowels the following consonants have most to be considered, and in initial consonants the following vowels. Mr. Curtis's calculations respecting the first letters of surnames are of much value. He used the commercial lists of the Post Office London Directory, and compared them with Liverpool, Hull, Manchester, Sheffield, Birmingham, and Bristol directories, and with three lists of different assurance companies; and after making his calculations from nearly 233,000 surnames, he found the total average very similar in its result. Mr. William Davis made similar calculations from the Clergy List, which came out much the same. These he contributed to Notes and Queries, [23] and subsequently he made a further calculation from French names. [24]

[ [23] 2nd S., vi. 496.

[ [24] 3rd S., iv. 371.

I have united these results in one table as follows:

MR. CURTIS. CLERGY LIST. FRENCH NAMES.
A 3·1 3·1 2·9
B 10·9 11·3 11·5
C 8·5 7·9 9·2
D 4·3 4·7 10·7
E 2·4 2·5 0·9
F 3·6 3·1 3·9
G 5·1 4·6 7·4
H 8·6 9·3 3·5
I, J 3·2 3·5 2·4
K 2·0 1·8 6·4
L 4·7 4·3 10·8
M 6·7 6·9 8·8
N 2·0 1·6 1·2
O 1·0 1·1 0·6
P 5·9 6·1 6·7
Q 0·2 0·0 0·3
R 4·6 4·4 5·3
S 9·7 7·7 4·3
T 4·0 4·4 3·3
U, V 1·0 1·3 3·2
W 7·9 8·3 0·8
X 0·0 0·0 0·0
Y 0·5 0·4 0·1
Z 0·1 0·0 0·0

MR. CURTIS.CLERGY LIST.FRENCH NAMES.
A3·13·12·9
B10·911·311·5
C8·57·99·2
D4·34·710·7
E2·42·50·9
F3·63·13·9
G5·14·67·4
H8·69·33·5
I, J3·23·52·4
K2·01·86·4
L4·74·310·8
M6·76·98·8
N2·01·61·2
O1·01·10·6
P5·96·16·7
Q0·20·00·3
R4·64·45·3
S9·77·74·3
T4·04·43·3
U, V1·01·33·2
W7·98·30·8
X0·00·00·0
Y0·50·40·1
Z0·10·00·0

It will be noticed that B is strongest in all three, and C is fairly equal. S is smaller in French names, but probably would be much larger in German names. H and W are also much smaller in French, while D and L are much larger. The preponderance of the latter letters is of course caused by the large number of names beginning with De and La.