IBM 604

From: Jerome Fine <jhfine_at_idirect.com>
Date: Tue Jul 3 08:39:10 2001

>John Allain wrote:

> You suggest a lot of responses here, so I'm a little shy
> to respond with only this. BTW nice history lesson.

Jerome Fine replies:

Well, at least one person took the trouble to respond. Thank you.
I don't mind taking 30 minutes (I type slowly) to provide some
historical perspective if someone is actually listening.

> > - could a Frieden do a division?
> Yes, on order of seconds. 2 secs for an extremely simple problem
> and 10 secs for a very complex one. Required use of a special third
> register. The algorithm was a simple one. More to the largest digit
> and subtract down the number until zero was reached. Every time
> zero is crossed, one time the divisor was added back and the next
> lower position of digit was then used. This process was painfully
> obvious just from the noise it made.

It certainly seemed MUCH longer. But I don't ever remember using
one myself to solve any actual problems. I suspect that by the time
a Frieden was possible (I do seem to remember they were larger
than a PC desktop box is today, probably larger than an old XT desktop
box - although probably smaller than a BA23 in its shell), shortly after
that, handheld calculators became possible.

> Was this one of the nonbinary "biquinary" machines?
> I guess one of the determining factors between computer
> and calculator.

You have the correct idea, but I think it was called "duo-quinary"
or 2 * "five states" for each character of the ten characters or
decimal digits in each word. But I can't remember how "five states"
were achieved. But that does mean that simple binary could be
replaced if a simple way of using such an ability were ever considered.
I suspect that with current concepts of RAM and having only an
OFF or ON state (simple binary), we have mostly forgotten that
other levels of representation were even possible, let alone used.

Of course, a punched card was the perfect example - 12 holes
per each of 80 columns with the lower 10 being the decimal
portion and the upper three zone holes able to extend the
range into alphabetic characters when two holes were punched
on the card. The big problem with such equipment was the
lack of parity checking, although relatively few errors ever
occurred percentage wise - although the absolute number
was obviously very large. The IBM keypunch almost always
produced a clean hold with no chad at all when it was adjusted
correctly. And with a correctly adjusted 029, it was even
possible to "add" or "remove" a column from the current card
by holding the correct drum - if someone did not touch type
so that the rest of the incorrect card could be repeated -
although I usually did a repeat of the new card just in case
the holes had not been punched in their exact locations after
the "unusual" operation had been done - the 029 was not
as particular whereas a card reader would require almost
perfect registration of the holes.

Sincerely yours,

Jerome Fine
Received on Tue Jul 03 2001 - 08:39:10 BST