On Jan 29, 2007, at 1:45 PM, Geoffrey T. Fong wrote:
> At 7:29 PM +0200 1/29/07, johnbaxterlists
mac.com wrote:
>> On Jan 28, 2007, at 8:40 PM, David Ross wrote:
>>
>>> Yep. I'm amazed at what gets patented these days.
>>
>> It's not just these days. National Cash Register (before the name
>> change to NCR) managed to patent gravity in the late 1950s.
>
> Hmm--sounds like an urban IP legend to me. I just Googled '"National
> Cash Register" gravity patent' and I don't see anything there. Can
> you provide a citation--or the patent text from the PTO?
You're looking for patents related to NCR's Card Random Access Memory.
You take a deck of longish, narrowish magnetic cards. You place a
series of notches in the "top" edge (one of the narrow edges). The
notches encode the binary address of the card (0 to 255 in the
original; 0 to 384 in a later model). The form of the notches fits a
series of horizontal, independently rotatable shaped rods over which
the deck of 256 (384) cards slides, with the cards hanging down. You
also have a pair of gating rods which hold all the cards regardless
of selector rod position.
Below the deck of cards you have a spinning drum (hollow, with
perforated surface at significant negative air pressure when a card
is present). To access the desired card, driver software sets the
selection rods to that card's address, and moves the gating rods out
of the way. Gravity causes the addressed card to fall, and guides
cause it to wrap around the drum, whereupon you have what is
effectively a small drum memory device. To read another card you
repeat the process for the new card, with the additonal process of
releasing the negative pressure in the drum while opening a gate
which allows the card to fly off the drum and through guides to where
it reloads onto the rods (from the rods' ends, with a magic loader
plate to push it into place).
RCA, to work around NCR's patents when building their "RACE" (I don't
remember what that one stood for), effectively turned the mechanism
upside down, with the selection rods on the bottom (as I had
explained to me, but on its side seems to make more sense), and used
other means than gravity to move the selected card to the drum.
I contend that RCA's reaction justifies my claim that NCR patented
gravity. (At least to the extent that Paul Heckel patented the XOR
instruction as applied to non-destructively modifying pixels on a
screen.)
IBM's machine was the "Data Cell". Think of an old 45 RPM jukebox
with the records stored around the playing mechanism. Replace the
records with...magnetic cards. Pick the desired card out of the
collection with a complex manipulator arm and wrap it around the
drum. (I think they were put back using momentum, as in the other
devices.)
The IBM Data Cell was probably the best of the machines, except for
one small problem: they required about an hour of Customer Engineer
PM time per hour of operation to keep them tuned up. Without that,
you had disaster sooner or later, usually sooner.
I never saw a RACE. I have operated CRAM many times. One of CRAM's
problems was the "screamer" or double drop (two cards at once).
There wasn't room around the drum for two cards--too thin a space.
So the machine uttered a rather loud scream as the drop slowed to as
stop due to the drag caused by destroying the cards. (Destruction
was pretty complete.)
One of NCR's salesmen was escorting a prospect into our data center
when one of the CRAMs executed a screamer. He turned to his prospect
and said, "Now you see, the machine has detected a bad card and
instantly destroyed it."
(The same salesman, in Dayton this time, escorted a group of
prospects into one of NCR's *big* centers, only to find that--despite
the appointment for the tour--they had torn up most of the tiles in
the walkways and were busily recabling everything. He said to that
group, "The machines work fine. We can't get this d... floor to work.")
All of these beasts gave way to RAMAC's descendants, but it took a
while to get disk* costs below these (especially that of a fully-
loaded Data Cell, which was quite large in capacity).
We had a running joke around NCR in the later 1960s because of
alternate spellings of "disk" (or "disc" or the pseudo?-French
"disque"). In non-formal papers, we would write disckque, to cover
the bases. The computer industry as a whole had trouble with this
before settling on disk.
--John the Sequipedalian*
* defined as "having a tendency to use words like Sequipedalian"
--
John W Baxter
jwbaxter
mac.com
![[F]](/Images/i7/site.gif){kind=small}
TidBITS 
TidBITS 