oper order. For the
same reason, as much of the counting and tracing should be done in the
plain impressions as it is possible to do.

If there is any doubt as to which of two or more classifications
should be assigned to a given pattern, it is given the preferred
classification and reference searches are conducted in all other
possible classifications. For example, if on a print with the
preferred classification

1 A
----
1 Aa

it is questionable whether the left middle finger should be a plain
arch, a tented arch, or a radial loop, the print is searched in the

1 A
----
1 Aa

group, and reference searches are conducted in the

1 A
----
1 At

and

1 A
----
1 Ar

groups. For further illustration, a print is given a preferred primary
classification of

1,
-
1

although the ridge detail on the right thumb is so formed as to
resemble a whorl. The search is completed first in the preferred

1
-
1

primary classification and a reference search is then conducted in the

1
--
17

primary.

All ridge counts that are "line counts," i.e., when one more or one
less count would change the designation of the loop from I to O or
from S to M, etc., must be searched in both groups. For example, in a
print classified

16 M 1 U III 10,
---------------
M 1 U III

if the ridge count of the right middle finger is 10 and the count in
the right thumb is 16 (as indicated by the key), the print would be
searched first as classified, then reference searches would be
conducted in the following groups:

M 1 U IOI, L 1 U III, L 1 U IOI
--------- --------- and ---------
M 1 U III M 1 U III M 1 U III

When there is doubt concerning the tracing of a whorl, it should be
treated in the same fashion. For example, if in the classification

O 5 U
------
I 17 U

doubt existed as to whether the tracing of the right thumb might not
be a meeting tracing, the print would be searched as classified, and a
reference search would be conducted in

M 5 U.
------
I 17 U

If there is no doubt concerning the ridge count used for the final, it
is enough to search out of the group only those prints containing a
final within 2 ridge counts on each side of the final on the print
being searched. When, however, there is doubt concerning the ridge
count of the final, the print should be searched 2 ridge counts on
each side of the two extremes of possibility. For example, if it were
possibl