Kilburn 1947 Report Cover Notes
Covering Notes for Tom Kilburn's 1947 report to TRE:
A Storage System for Use with Binary Digital Computing Machines
This Report was the key document written by Tom Kilburn to
describe the work done in 1947 to investigate the possibility
of using a conventional Cathode Ray Tube to provide a storage
system for electronic computers.
The 1947 Report was written by Kilburn for TRE, as a report
on his first year of secondment from TRE, and to enable a second year of secondment.
As Simon Lavington notes in the Reference
to Kilburn's thesis in his book
"A History of Manchester Computers", the Report was "a personal
report to TRE to obtain a second
year of secondment to the University of Manchester. Kilburn
has stated that whilst he was writing it, requests for copies
came in. Twenty copies were therefore made, and the document
was circulated as an Internal Report by the Department of
Electrotechnics, University of Manchester, dated 1st December
1947. Due to demand, a further 30 copies were made. It is
known that several copies were taken to the USA by Douglas
Hartree (Cambridge), Harold Huskey (SWAC at UCLA) and
A M Uttley (TRE) in the Spring of 1948."
Kilburn used the same title for his PhD
thesis (in 1948), and the title is used yet again for the first detailed
paper published on the Manchester C.R.T. storage mechanism (in 1949),
i.e.
Williams, F C and Kilburn, T :
A Storage System for Use with Binary Digital Computing
Machines, Proc. IEE, Vol. 96, part 2, No 30, 1949
This 1949 Paper is closely based on the 1947 Report.
It adds very little to the report. Indeed it
removes the material on both the less and more
detailed "hypothetical machines" of the report, which are
given to illustrate how the storage system described might be
used in practice. (Of course by the time the paper was published
computers using the storage now existed!) This material is
interesting in that it shows an understanding of what a
small computer using the C.R.T. store might look like
that is significantly different from what Tom actually designed
a few months later.
(In particular, he imagines that each instruction will have its
own dedicated register, and the instruction code simply comprises
two addresses and a "direction" bit, the register address
implying the instruction. Also the act of bringing down the next
instruction is treated as an instruction, hence the early reference
to a computer using his C.R.T. taking 600 microsecs. per
instruction, rather than the 1200 one would expect knowing
the Baby!)
Note that in general the individual pages provided which show
graphical material (e.g. a diagram) are less than 100KB in size.
Exceptions are detailed on the page listing them all.
On my browsers at least, the diagrams will come up scaled
down as necessary to fit in the window, and you can increase the
size by increasing the window size, or by clicking as appropriate
to the browser to bring the diagram up to full size, when you
can then scroll around it.
The material provided in the pages associated with 1947 Report is
as follows :
- The text of a transcription of the 1947 Report.
- A summary of the editing that has been done in the transcription.
- Inexpensive scans are provided of pages 6, 7, 18, and 19 if you want to get a feel
of the original Report. A 4.5MB version of page 19
exists if you want to get a closer feel!
- A
summary is provided of the differences between the 1947 Report
and the 1949 Paper.
- The diagrams of the Report are tied in
at the back of the Report, done as blueprints in a wide variety of
sizes. These have been converted into conventional black on white,
with a certain amount of cleaning up.
To get a feel for the originals,
see 2.4 and 3.6 (but the blue is a bit too
bright!).
The blueprints
were prepared by Joe McCormick.
- The set of diagrams and photographs from the Report are
stored in separate pages (in black and white) and are referenced
from the transcription text in a conventional manner.
But an index of the diagrams
is given on a separate page
to help anyone who wishes to open a separate window to study a
diagram with the associated text still visible. (E.g., open a second
window with a copy of the diagrams index URL and click on
the appropriate diagram number in the index to see the diagram; go
BACK to the index when you have finished, ready for the next one.)
- The modern standard paragraphing convention is used,
whereas the Report indents at a paragraph, with no blank line.
- Hyperlinks have been superimposed on numbers in the text to
provide linking to sections of the text, figures and
references in a conventional manner.
- The Report uses footnotes, mostly to give the references, but once
or twice to make insertions in the text to achieve corrections.
Each has been replaced in situ by the footnote, without brackets
where they are insertions, and in brackets for the references.
E.g. :
"For these reasons the binary system of numbers has become popular
in recent plans for electronic digital computing
machines1, although in the past the decimal system has been
used2." becomes
"For these reasons the binary system of numbers has become popular
in recent plans for electronic digital computing machines
(bibliographies 1, 2), although in the past the decimal system has been used
(bibliography 3).".
- Most of the mathematical terms which contained division in the Report
used three lines, with a line separating dividend and divisor. They
have been replaced by dividend "/" divisor, all on one line. In two
cases a pair of brackets have been added to specify the correct
binding.
- Various spaces have been removed to achieve a consistent usage
throughout the Report, e.g. within a term in a formula, and
between a number and its unit (e.g. 6.3mms.)
- The Report's usage of "o" instead of "0" has been retained for
zero subscripts and superscripts!
- The edited version consistently uses a capital S and F for Section
and Figure references, and I have consistently used the hyphen in
"Y-shift" rather than the occasional "Y shift".
- There are a number of minor typographical errors in the original,
including missing "small words" (e.g. prepositions and articles)
and a couple of incorrect references. Where I am confident of
the error, I have corrected it (usually confirming it from the
corresponding passage of the 1949 Paper).
- I have in general not tried to improve the punctuation,
e.g. I have kept to "C.R.T.'s" rather than "C.R.T.s" as the
plural of "C.R.T." !
- The Report has some local inconsistencies in the use of
subscripts, especially w.r.t. elements in diagrams. I have
made these consistent with the local majority usage
(i.e. to use subscripts). But in section 8, where there
are a lot of such references to valves (e.g. V2) and diodes
(e.g. D3) and none use subscripts, I have
left them without subscripts. (The 1949 Paper consistently uses
subscripts throughout.)
- There is a particular passage,
Section 4.1, where a lot of
references are made to the triodes Tn'. The 1949 Paper corrects these to T'n (in both text and diagram) but I
have left them so that they remain consistent with the Report
diagram (see Figure 4.4).
Except as implied by the detailed differences given below, the 1949 Paper is mostly a verbatim copy of the 1947
Report, and the diagrams are nearly all identical in content and in
the same sequence.
- Section 1.3, 1.4 and 1.5 have been reordered, with some rewriting,
and in particular the references to the schematic diagram of a
hypothetical machine have been removed (as indeed has the
diagram, Fig 1.2).
- Section 4.1, which contains detailed references to two circuit
diagrams, has been rewritten at a much higher level, with most of
the detailed description put in the Paper's Appendix, which is
the equivalent of Section 8 of the Report.
- Section 6 has been removed. This relates in more detail how the
C.R.T. store would be used in a computer (with a more sophisticated
schematic diagram 6.1 -- also removed).
- A short Acknowledgements and Conclusions section is added in the Paper.
- Section 8 of the report becomes an Appendix in the paper, with
the material of 8.3 and 8.4 removed, and the material from Section 4.1 added.
- The diagrams of the Paper essentially comprise the diagrams in
the Report, minus those relating to the removed material
(Figures 1.2, 4.3, 5.5, 6.1, 6.2, 6.3, 8.4, 8.5 and 8.6). The
diagrams that remain are in general identical in content
and relative position, but altered in presentation.
Some of the redone circuit diagrams are unpleasantly cluttered.
There are two with minor alterations
in content, 3.3 and 8.1,
and one diagram with substantial alterations, 8.2. Also, in the Report,
Figure 4.5 is spread over two diagrams (both labelled 4.5),
which I have labelled 4.5a and 4.5b behind the scenes,
with the hyperlink in the text going to the appropriate one
for the context. But in the Paper 4.5a is kept in its original
position and 4.5b is moved to the
appendix (along with 4.4).