Danny Burstein wrote:
> The counters (one for each phone line) simply clicked off as needed. I
> have no idea how they got the feedback to "know" whether to just click
> once for a local, untimed, call -- or whether to click twice on start,
> once every three minutes (or some other arrangement ...) depending on
> distance.
The Eng & Sci book and early writeup on panel describes how the
counters were incremented for local calls. Under operators, she
pressed a key. For automation, certain signals were sent along the
sleeve (as opposed to tip and ring) at disconnect time if the call was
answered for more than 2 seconds. Indeed, in switching lots of
signals were sent that way, from small/high pos/neg currents to AC
signals. Automation knew at the start not to increment for "official"
calls.
It is relatively easy to increment once per call. To increment for
suburban calls, the system would need (a) a timer and way to increment
per period, and (b) a machine lookup table to know the base increment.
From the readings I've gotten so far, it appears the conversion of
suburban calls to message units (with direct dialing instead of
operator handling) happened after WW II. Before that such short calls
were operator handled. NY Telephone was quite emphatic that dial
wouldn't reduce operator staff and it didn't since so many calls still
requird operator assistance and traffic was growing so rapidly. In
NYC, calling between boroughs or a long distance within a borough was
toll and an operator was required. Panel routed such calls to
intercept.
> NY Tel had a camera that literally was pushed against a (dozen at a
> time?) counters and the numbers were then processed by someone or
> another and added to the bill.
That was part of the panel system.
In the 1920s and 1930s I can't imagine the tedious job of posting all
the camera readings to customer accounts then subtracting last month
from this month. There was a sliding scale of message unit cost--the
unit cost went down for large consumption. Hopefully in the 1940s
they used IBM machines which were ideal for that application. (the
"Mark sense" cards used as toll tickets were fed directly into IBM
machines).
hancock4@bbs.cpcn.com