wrote:
>> However, in the 1950s, IBM developed card-to-card directly without
>> paper tape and "over AT&T lines". Modems were developed to take good
>> advtg of the available bandwidth (about 1200 baud). Undoubtedly the
>> equipment and implementation was developed in close cooperation with >> AT&T.
> This was the IBM "Card-to-card" transceiver. I don't know when they
> first came out, but the Army started implementing them in a nationwide
> network in September of 1956.
H'mmm ... was this implemented with an 026 (or something that looked a lot like an 026), with a narrow cabinet bolted on the right side? I never worked with one, but I have a recollection of seeing such a beastie in the Atlanta datacenter in the late 1960s when the sysprogs from my PPOE went there in preparation for the installation of some big IBM boxes at our shop.
> I was wondering if the modems in that application were supplied by IBM
>> (who appears to have developed the technology) or by AT&T. My
>> understanding that AT&T's "Dataset" modem-telephones didn't come out
>> until the 1960s.
c /Dataset/DataPhone (tm)/
> Comments by anyone familiar with pre-1960 data communications would be
>> greatly appreciated.
> I believe they used 4-wire leased lines, with data access arrangement
> boxes provided by Ma Bell. So the signals going into the big grey box
> next to the reader/punch were analogue. I don't recall what the
> transmission rate was, but they sent EBCDIC directly without any
> translation to a 5-channel code and no added headers.
The only "data access arrangement" I know of was the stupid box that the FCC mandated had to be used to attach non-Bell modems to the dial network for a (thankfully brief) time post-Carterphone. Users who used a leased circuit could (within limits) attach just about anything they pleased.
The circuits you're referring to were probably 1009 (direct copper) assuming that my memory hasn't had too much bitrot since then. I do recall sometimes having to play games with the tariffs and (for reasons I can't remember) sometimes having to get a 3002 circuit instead, then go around and disconnect the conditioning boxes so that the line would act like a 1009.
Joe Morris