make the _B_-board entirely separate
from the _A_-board, although the general characteristics of construction
remain the same. The reason for separate _A_- and _B_-switchboards in
large exchanges is to provide for independent growth of each without the
growth of either interfering with the other.

A portion of an incoming trunk, or _B_-board, is shown in Fig. 378. The
multiple is as usual, and, of course, there are no outgoing trunk jacks
nor regular cord pairs. Instead the key and plug shelves are provided
with the incoming-trunk plug equipments, thirty of these being about the
usual quota assigned to each operator's position.

In multi-office exchanges, employing many central offices, such, for
instance, as those in New York or Chicago, it is frequently found that
nearly all of the calls that originate in one office are for subscribers
whose lines terminate in some other office. In other words, the number
of calls that have to be trunked to other offices is greatly in excess
of the number of calls that may be handled through the multiple of the
_A_-board in which they originate. It is not infrequent to have the
percentage of trunked calls run as high as 75 per cent of the total
number of calls originating in any one office, and in some of the
offices in the larger cities this percentage runs higher than 90 per
cent.

[Illustration: Fig. 378. Section of Trunk Switchboard]

[Illustration: Fig. 379. Section of Partial Multiple Switchboard]

This fact has brought up for consideration the problem as to whether,
when the nature of the traffic is such that only a very small portion of
the calls can be handled in the office where they originate, it is worth
while to employ the multiple terminals for the subscribers' lines on the
_A_-boards. In other words, if so great a proportion as 90 per cent of
the calls have to be trunked any way, is it worth while to provide the
great expense of a full multiple on all the sections of the _A_-board in
order to make it possible to handle the remaining 10 per cent of the
calls directly by the _A_-operators?

As a result of this consideration it has been generally conceded that
where such a very great percentage of trunking was necessary, the full
multiple of the subscribers' lines on each section was not warranted,
and what is known as the partial multiple board has come into existence
in large manual exchanges. In these the regular subscribers' multiple is
entirely omitted from the