COMMENT ⊗   VALID 00006 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	%\magnification=\magstephalf
C00008 00003	\stanford
C00072 00004	\def\today{\ifcase\month\or
C00077 00005	Let's consider {business letters} first. Suppose that you want \TeX\ to
C00092 00006	\beginchapter Appendix E. Example Formats
C00108 ENDMK
C⊗;
%\magnification=\magstephalf
\magnification=\magstep1
\font\smallheadfont=amr8 
\font\largeheadfont=amdunh10 
%\font\ninerm=amr9 
\font\seal=stan70
\font\Dunh=amdunh10
\def\today{\ifcase\month\or
  January\or February\or March\or April\or May\or June\or
  July\or August\or September\or October\or November\or December\fi
  \space\number\day, \number\year}

\raggedbottom
\interlinepenalty=1000
\hsize=6.25truein
%\voffset=24pt
%\advance\vsize by-\voffset
\parindent=0pt
\parskip=0pt
\nopagenumbers
\footline={\ifnum\pageno>1
  \smallheadfont \addressee\hfil\today\qquad \folio
  \else\hfil\fi}

\def\beginlinemode{\endmode
  \begingroup\obeylines\def\endmode{\par\endgroup}}
\def\beginparmode{\endmode
  \begingroup\parskip=\medskipamount \def\endmode{\par\endgroup}}
\let\endmode=\par
\def\endletter{\endmode\vfill\supereject}

\newdimen\longindentation \longindentation=4truein
\newbox\theaddress
\def\address{\beginlinemode\getaddress}
{\obeylines\gdef\getaddress #1
  #2
  {#1\gdef\addressee{#2}%
    \global\setbox\theaddress=\vbox\bgroup\raggedright%
     \hsize=3.25truein\everypar{\hangindent2em}#2
%    \hsize=\longindentation \everypar{\hangindent2em}#2
    \def\endmode{\egroup\endgroup \copy\theaddress \bigskip}}}

\def\body{\beginparmode}
\def\annotations{\beginlinemode\def\par{\endgraf\nobreak}\obeylines\par}
\def\ps{\beginparmode\nobreak
  \interlinepenalty5000\def\par{\endgraf\penalty5000}}

\def\up#1{\leavevmode \raise.16ex\hbox{#1}}

\def\personal{\pageno=1
  \def\sendingaddress{Arthur L. Samuel\par
    The Sequoias\par
    501 Portola Road, box 8214\par
    Portola Valley, CA 94025\par
    \up[415\up]\thinspace 424-4233\par}
  \def\returnaddress{Arthur L. Samuel\par
    The Sequoias\par
    501 Portola Road, box 8214\par
    Portola Valley, CA 94025 USA}
\centerline{Arthur L. Samuel}
\centerline{The Sequoias}
\centerline{501 Portola Rd. box 8214}
\centerline{Portola Valley, CA 94025}
    \bigskip\bigskip\hfill\today\bigskip}

\def\stanford{\pageno=1
  \def\returnaddress{Arthur L.~Samuel\par
    Computer Science Department\par
    Stanford University\par
    Stanford CA 94305 USA}
{\null\vskip-10pt
\vbox{\hskip-50pt$\vcenter{\rlap{\seal T}}$\hskip48pt}
\vbox{\null\vskip-66pt\Dunh\parskip=0pt\baselineskip=18pt\obeylines\obeyspaces
   Computer Science Department\par
   Stanford University\par
   Stanford, California 94305}}
    \bigskip\bigskip\hfill\today\bigskip}

%\def\makelabel{\endletter\hbox{\vrule
%    \vbox{\hrule \kern6truept
%      \hbox{\kern6truept\vbox to 3truein{\hsize=\longindentation
%          \smallheadfont\baselineskip9truept\returnaddress
%          \vfill\moveright 3truein\copy\theaddress\vfill}%
%        \kern6truept}\kern6truept\hrule}\vrule}
%  \vfill\eject}

\def\makelabel{\endletter\hbox%
{\vrule\vbox{\hrule \kern6truept
      \hbox{\kern6truept\vbox{\hsize=2truein
    \smallheadfont\baselineskip10truept\returnaddress}}\smallskip\hrule}\vrule}
\medskip
\hbox{\vrule\vbox{\hrule \kern6truept
      \hbox{\kern6truept\vbox{\hsize=3truein
     \copy\theaddress}}\smallskip\hrule}\vrule}\eject}

\def\als{\beginlinemode\nobreak\bigskip
\hskip4truein Sincerely yours,
    \nobreak\bigskip\bigskip\bigskip % space for signature
\hskip4truein Arthur L. Samuel
\bigskip}
\def\fin{\makelabel\end}

\stanford
\address


\stanford
\address
Profeessor P. Masani
Department of Mathematocs and Statistics
University of Pittsburgh
Pittsburgh, PA  15260

\body
Dear Professor Masani:

The long awaited material is enclosed.
I really must apologize for my failure to handle this matter more
promptly, but I have been under unusual pressure to finish a much needed,
rather large, computer program.

As you suggested, I have written the added text with the [53g],
[53i], [53j], [54c], [60b] order in mind. I am afraid that I have not
spent enough time on this endeavor but here it is in any case.

I note one possible error: In my comments on [53g], I refered to [60d]
while the final paper is, I believe, the one otherwise refered to as
[60b].

\als
\fin

\centerline{Comments on[53i], [53j], and [54c]}

[53i] In this paper, which was presented in December 1952,
Professor Wiener did make some predictiona as to the
future of automatic machinery and in specific on the electronic brain and
the next industrial revolution.  He rightly took exception to the
extravagent claims that were being bandied about concerning the future use
of computers, quoting as a rather extreme example some remarke by Bertrand
Russell.  At the same time, Professor Wiener showed remarkable perspicuity
in calling attention to the likelihood that the computer would not
necessarily simply replace man in the performance of some specific task
but that the task itself would be redefined to better adapt it to the
computer's capabilities.

[53j] The theme of The Machine as Threat and Promise was again covered in some
detail in a popularized article that appeared in the St. Louis Post-Dispatch
on December 13, 1953.  In this article, Professor Wiener's principal concern
was to dissuade the public from a belief that the transition to an
automated economy would or perhaps even could be made without serious
dislocations and perhaps some changes in our social structure.  On the whole,
however, he noted ``that the prosperity which is in sight as a result of 
automatiation [sic] is ours to grasp if we really wish to grasp it.''

[54c] Following up on the earlier paper in the same paper, Professor
Wiener, this time with the collaboration of Professor Donald Pierse
Campbell, published an article on ``Automatic Factories---Fact and Fiction
in the St. Louis Post Dispatch roughly one year later.  In it he tried to
draw a clear distinction between the ultimate, long term, desirable
progress toward an automated society and the quick fix envisioned by many
people and he warned that ``Every new invention has gone through a
speculative period, when bubbles have come to the surface, and bubbles
have burst.''  Viewed in retrospect, some of Professor Wiener's concerns
were unnecessary, or perhaps it was his timely warnings that saved us from
some of the troubles that he forsaw.  Progress has not been always
continuous, we have seen some bubbles and there may be some yet to burst,
but the progress has been spectacular and the miniturization of the
computer chip has lead to progress that Professor Wiener was quite unable
to forsee.





\stanford
\address
The Director
The Computer Museum 
One Iron Way
Marlboro, MA 01752

\body
Dear Sir:

Your membership folder states that ``Exhibitions will offer the visitors a
chance to play the first computer game, \dots''. How do you propose to do
this, since all of the early games were programmed for computers that are
no longer available?

There is also the problem of identifying the ``first'' game.  Most of the
early computers were programmed to play pseudo-games such as tic-tac-toe
and nim (which are not real games since their solutions are known). Even
admitting these games, it is hard to say, at this late date, what was the
first game, hard to duplicate it, even harder to resurect or duplicate the
machine on which it was played, and quite impractical to maintain the
machine and keep it in vacuum tubes.

If pseudo-games are to be admitted, then you might consider resurecting
the Nimrod machine which was featured at the Festival of Britain in 1951 
and which used only 480 vacuum tubes,
although this machine was not a true computer (but then neither were some
of the more recent commercial, so-called computer games).

If you are to admit machines that were not true computers (with all
functions that Babbage proposed for his Analytical Engine) then you would
have to include Torres Quevedo's king and rook mating machine built in
1890. If you exclude such machines then game played on the original ENIAC
and the Harvard Mark 1 cannot be included.

One of the very early true computers (having a conditional transfer), if
not the very first, was the IBM-SSEC which was put into operation in early
1948. I joined IBM in 1949 and I was unaware of any games having ever been
played on this machine although it is hard to believe that some of the
early users of the machine did not try out a game or two on it or that
some games were not used during the initial testing period.

Excluding pseudo-games and games on which no records were kept, I believe
the credit for programming the first true computer game (that is, a true
game on a true computer) should probably go to Christopher Strachey for
writing a draughts (checkers) program in 1950. Several simplified chess games
came along at about this same time, not allowing certain pieces, on a 6 by
6 board, not allowing castling, not differentiating between checking and a
checkmate, etc..

If you are to require that the program had to be able to play a complete
game (Strachey's program did not play properly during the end game) and that it
had to have been made available to others, then I wrote a
complete playable checker game for the first IBM-701 computer circa 1951
(written in numerical machine code).

This program (when disassembled into assembly code and reassembled) was
demonstrated on television a number of times and played by many people
including Walter Hellman, the then world checker champion.  I also
distributed copies of the program and it was put into operation on quite a
few different computers.  In the more recent past, I reprogrammed checkers
for one of the early commercial computer-game machines, but it would be
quite false to call this modern version the same program.

I certainly do not claim to have written the `first computer game', but my
work does set an early date that must be bettered for any game to be
called the `first computer game'.

I recall seeing several other early games during the 1950's, mill, hex,
biliards, go moku, and a space war game, to name a few, but most of these
were written rather late in the 1950's when computers came into more general use
and when scope displays became commonly available.

Which of these do you believe to be the ``first computer game''? It would be
quite dishonest to bill a demostration game as the ``the first computer game''
if it had been reprogrammed to run on a present day computer.

\als
\fin

\stanford
\address
Professor P. Masani
Department of Mathematics and Statistics
University of Pittsburgh
Pitttsburg, PA 15260

\body
Dear Professor Masani:

Thank you for making a much needed correction to my Comments.

My unthinking original word choice was prompted, I fear, by my recalling
that, in truth, Professor Wiener and I were quite unable to come to any
agreement when I attempted to discuss this matter with him on several different
occasions.

The word `controversy' is certainly more in keeping with an objective
evaluation of Professor Wiener's many many contributions.

One might even go so far as to use the word `disagreement' although perhaps
`controversy' is the better word since it implies a continuing disagrement.

\als
\fin

\address
Professor Pierre A. MacKay
Department of Computer Science
University of Washington
FR-35
Seattle, WA 98195

\body 

Dear Pierre:

Thanks for the information on printers.

Here is an Information Sheet that we have prepared to have something to
send to people who write in for information while knowing practically
nothing about \TeX.  If you see something that is grievously in error, do
let me know.

You are welcome to use this in part or in whole if you find a use for it.

\als
\fin

\address
Stanford Help Center
Galvez House

\body

My comments on the Retirement Workshop on January 18th do not quite fit
into the format on your workshop evaluation sheet, so I am writing this letter.
Also my hand writing is a bit hard to read (I may be getting a bit old).

In the first place the workshop was not helpful to me although it met my
expectations.  I was not surprised by what was said, and in fact Dr. Gulevich
gave a very clear exposition of the
troubles that face older people.

The trouble is that most older people already know about these troubles.
There may be some `misery loves company' aspects in being told
that these troubles are common to all old people but I for one do not like
being reminded that there is nothing that I can do about it.  Particularly
when there is, and that is to not think about my troubles, to keep busy,
and to not retire.

I do not thing that active people should ever retire.  Oh, of course, they
may have to reduce the level of their activities as they get older and
they may even have to change jobs because of inhuman retirement rules, or
if their physical stamina is no longer up to the requirements of their
earlier work but they should keep working. When people retire they cease
to carry their share of the world's work and I for one would not be happy
if I did not continue to contribute as long as I am physically able to do so.

So, I thing that a workshop on retirement should be devoted to a
consideration of the ways in which an older person can thwart the all too
prevalent compulsory retirement syndrome. It may well be that Stanford might
not look with favor on this activity on the part of the Help Center, but it
would be interesting to see what would happen if you gave it a try.

\als
\fin

Mr. Derek Oldbury
4 Farm Close
Kingskerswell
Newton Abbot
Devon
ENGLAND

\body
Dear Mr. Oldbury:

I am very glad indeed to hear that you have become interested in
computing.  You should be able to generate a very much better checker
playing program that I was able to do, all because of your vastly greater
knowledge of checkers.

I am sorry to have been so long in answering your letter.
I have started to answer it on several occasions and then the pressure
of other work has forced me to put it aside.  My problem is that I have
not been working on checkers for some years and I am finding it very
hard to order my thoughts to make any worth while suggestions that
might aid you in your work.

Sad to relate, very little computer work has been dome on checkers since
the 1960's.  Everyone seems to be interested in chess these days. Even
here, the emphasis has been mainly on brute force computing and nothing
worth mentioning has been done to apply learning principles.  Also, the
entire field of Artificial Intelligence has been largely taken over by
workers whose primary interest is in Expert Systems. The intent is to
store man-derived information which is then partially digested and
regurgitated on demand.  Machine learning is a badly neglected field.
I am pleased to learn that you, at least, appreciated what I was trying to do.


I did work with one or two good checker players for a while in an attempt
to arrive at a better set of paramaters but I was signally unsuccessful.
In thinking back on this work I know at least one way in which I went
wrong. Perhaps this might be a worthwhile thing to mention.

What I was doing was to take some of the standard book games and ask a
checker master to list the reasons why each particular move was chosen.
I would then try to codify these responses into some sort of routine or set
of parameters that would capture the essence of the ideas involved and use these
in place of my earlier parameters.

I now realize that I was asking the wrong question.  It would have been
better to list all of the rejected moves at each point, excluding the
recommended move and any alternate recommended moves. I should then ask for
the reasons why each of these bad moves was rejected.  Not only would I
accumulate much more information (usually at least 4 bad moves for each
good move and frequently many more) but the information would also be much
more decisive and perhaps easier to codify for use by the computer. One
continuing problem is to reduce the amount of branching. If the computer
could sort out the very bad moves at once it could do a lot of forward
pruning, something that the checker master seems to do very well indeed
but that has been hard to do in a computer progrram.

As for the choice of a computer languages, I have never favored LISP
for checkers. The problem has always been one of computer speed and
not one of logical complexity.  The rules of checkers are really quite
simple and minimaxing is easy to do in any one of a dozen languages.  PASCAL
is a reasonably good choice.  Actually I have done most of my checker work
in an assembly language for speed reasons but if I were to get back to
checkers, I would now use a higher-level language.  Computers are
so much faster these days and compilers are much better than they were in
the 50's and 60's that there is no longer the need to use assembly coding.

You did not tell me the make of computer that you intend to buy.  I have an
IBM-PC at home and I have occasionally thought about writing a checker
program for it. As evidence of what can be done, I wrote a simple
checker program (without learning) for an early home game computer (called
the VideoBrain) that is now no longer on the market. The entire program was
tailored to fit into 8k bytes of ROM memory, it used a joy stick for
input and displayed the board on a TV. This program had a fairly large sale and
it played a surprisingly good game
although not nearly good enough to interest you.

I do hope that these ramblings may be of some use to you and I would very
much like to hear of your future progress.


\als
\fin

\address
Mr. Sanjoy Mahajan
121 Wilmar Drive
Pittsburgh, PA 15238
\body
Dear Mr. Mahajan:

In your letter of December 27, 1983 you asked if learning techniques had
been applied to chess or go.  The answer is really no although several people
have made starts at doing this.

Perhaps the best work on go was done by Jon Ryder.  He finally settled for
a program that played an amateur game of go without any learning.  The
trouble with chess as a learning vehicle is its complexity.  The program
(and the designer of the program) must spend a great deal of effort in
just handling these complexities.  In the case of go the rules are even
simplier than the rules for checkers but the game is very profound and the
branching factor is so very large that look-ahead is of limited use except
in certain situations, such as laddering, when one then has to look ahead
sometimes for twenty or more moves.

I am afraid that you do not quite understand what you are asking for when
you ask for a copy of my program. In the past when I made copies for
others, it was always on magnetic tape for use on other compatable 36-bit
computers from files that were then available on our computer. Now all of
this material is long since buried in a room-full of archival tapes. I
would have to find and load at least five different files (the program was
always compiled in several sections because of its size), printing all of
this material, something that I seldom did, because of its bulk,
and finally send you a three or four inch stack of print-out paper that
would be of no use to you.

Were I to start working on checkers again, I would certainly not bother to look at
my old program.  Instead, I would start all over and recode the game from
scratch, probably in PASCAL, although I might use WEB, the new language
that \TeX\ is written in, because of its superior documentation
capabilities.  With the availability of modern optimization compilers, the
speed advantages of using an assembly language are less important.

As for the differences between using a 32-bit word machine and a 36-bit
one, I can only say that when I coded the game for the VideoBrain, I found
that using an 8-bit byte was not nearly as much trouble as I had at first thought
it would be.

\als
\fin



Ms. Barbare Beeton
American Mathematical Society
P.O. Box 6248
Providence, RI 02940

\body
Dear Barbara:

I trust that you have received the revised First Grade \TeX\ by now.
  
You did such a good job on First Grade \TeX\ that I am anxious to see if you
might be willing to do the same thing to the enclosed information
sheet.  This will be reproduced on a single 11/17 inch sheet
to make a 4-page folder that can be sent out with an order-form insert.

This information will be mailed to people who know little or nothing about
\TeX\ and who phone or write us for information.  You might also find the
sheet usefull and you are welcome to use it as-is or modified as you see
fit.  I will be glad to provide an APS'ed copy if you want it.

The tables are quite incomplete and will be prettied up when all of the
desired information is at hand..  Can you supply some of the missing items?

\als
\fin
Dated Jan. 9,1984

\personal
\address
Mr. Randy Finch
532 West 111 St. \#36
New York, NY 10025

\body
Dear Randy:

This may not be quite as personal as a hand written letter, but it is a
bit more readable than my handwriting and I though you might like to see
the kind of letter that our computer produces.  

Every time that I fix
myself some tea, I will now have a ready reminder of your kindness in
remembering me this Christmas.  Thanks a lot.

It is about time for me to send you the enclosed check
so that you will be sure to have enough ready cash to cover your ticket. I
believe that \$2000 should be enough but do let me know if it is not.

I have decided not to come home by way of Rome.  It seems
to be just a bit too much travelling for me. After all, I have seen Margie
in Europe, not in Rome, it is true, but we did have three weeks together
last summer.

I hope that you have gotten your passport by now.  If it is slow in coming,
you should call the visa people and warn them that your application may be
a little late.  My passport came back in just a little more than a week
from the time that I mailed the application.

\als

\fin

\address
Travel Dynamics Inc.
att. Ms.~Harriet E. Harris
1290 Ave.~of the Americas
New York, NY 10104

\body
Dear Ms.~Harris:

I am enclosing a check for \$1691.00 to cover my round trip airplane fare for
the combined Stanford Alumni Indonesia College and Indian Ocean College trips.

I understand that my ticket will be for the following:  San Francisco to Hong
Kong on Singapore Airlines \#1 departing San Francisco at 10:45 PM on
Thursday March 1, 1984; Hong Kong to Denpasar on Garuda Indonesia Air \#875
departing Hong Kong at 3:00 PM on March 5th; Colombo to Singapore on
Singapore Airlines \#27 departing Colombo at 12:50 PM on April 7th; and
Singapore to San Francisco on Singapore Airlines \#2 departing Singapore
at 3:30 PM on April 8th.

\als
\fin 

\personal
\address
TRAV-ALL Visa Service
Att. Mr.~Ray Riddle
P.O. Box 523
Falls Church, VA 22046

\body
Dear Mr. Riddle:

I am enclosing my U.S. passport No. 050207452 issued Dec. 14, 1983, five
passport photographs and the visa application forms as supplied to me by
the Stanford Alumni Association, in connection with the Indian Ocean
College trip from March 15, 1984 to April 8, 1984.

Please note that I am also taking the Indonesia College trip (that preceeds the
Indian Ocean College trip) and that I will be leaving my home on March 1, 1984 or
possibly a few days earlier so that I can break my flight in Hawaii.

Should you wish to reach me by phone, I can be reached on (415)497-3330 during
the day (8:30-11:45AM, 1:00-4:00PM  PST) and on (415)424-4233 during the evenings.
These phones will not normally be answered at other times.

\als
\fin

Prevent Blindness 
Department MM
1208 Pioneer Building
St. Paul, Minnesota 55101-1592

\body
Gentlemen:

I am enclosing a check for one dollar to cover postage and mailing costs
for a copy of your Minnesota Adult Home Eye Screening, as described in a
recent article in Modern Maturity.

\als

\fin

\address
Mr. and Mrs. R. Mussen
40 Pearl Street
Essex Junction, VT 05452

\body
Dear Ron and Louise:

It was nice to hear from you. I am sorry that I did  not write you last Christmas,
but I simply did not write any one.

Bernice died in May of 1982.  She had had a coronary bypass
operation in January of that year and she never fully recovered. I miss
her greatly.  Fortunately, we moved into The Sequoias some years ago, so I
am not as much alone as I would be if I were living in a single family
home.

I was glad to hear that you are retiring, since you both seemed to desire
retirement so very much.  However, it does seem that Ron has plans to keep busy,
and it sounds like you also find plenty to do.

As for me, I am still working and I doubt that I will retire until failing
health actually forces me to do so.  I have now been at Stanford a little
longer than I was at IBM, 17 years at IBM and 17 years and a few months at
Stanford.  I have been unusually lucky in being able to get a job
after I retired from IBM and to hang on to it for
these many years. Yes, I still spend my days in front of a terminal, what
with a terminal in my office, another one in my study at home and an
IBM-PC at home as well.

In a sense, I suppose, I am retired, in that I am very much my own boss
and only work when I want to.  At my own insistance, I have been
gradually reducing my work week so that I am now officially working only
one fourth time although I still put in a 40 or 50 hour week when I am not
off travelling.

Bernice and I had gone around the world on the Rotterdam in 1971 and we
had planned to repeat this trip in 1982.  When Bernice first took sick, we
postponed the trip a year and after Bernice's death, Margaret and Donna
insisted that I simply must take
the trip alone. 

So I started this year off with a long trip. I took my IBM-PC with me
and I gave a series of lectures and demonstrations on the ship.  I had given a
series of lectures
on the Rotterdam in 1971 (that time on the game of Go), so this was
not a particularly new thing for me to do.  It did keep me busy and it was
a good way to get acquainted with the passengers, something that might
otherwise have been hard for me to do.

Then in May I went with Donna, her husband and their two children for a
trip on a wind-jammer in the Caribbean.  This was a bit rugged for a
person my age but I got along very well indeed and I
enjoyed it very much.

Finally, to top it all, I took a rather long European trip in July
and August with Margaret and Donna.  I was given an award that also
paid my expenses to a conference in Germany and I had been asked to give a
talk at the presentation ceremony.
The two girls wanted to be there to witness the
presentation.  Margaret is taking a year's leave from her job at Hartford
University to work on her PhD dissertation studying in Rome, and she was
already in Europe.  So the three of us combined a trip to Germany with a
boat trip down the Rhine, a visit to London and a boat trip along the
`Waterways of Western Europe'.

Having mentioned what Margaret is doing, I must say something about Donna.
She and her husband have been quite successful in writing computer texts.
Pasha had written several small texts that never sold very well and when
he started on a full size book, Donna complained
about his english and he challenged her to see if she could do better.
She had been doing the typing for him so this time instead of typing what
he had written she rewrote every thing in her own words.  The book was an
immediate success.  They have written a second book, following the same
pattern, are now writing a third one and have signeed a contract for a
fourth.

I do count myself fortunate to have such good daughters, and we have
become even more close since Bernice died.  I don't know how I could
managed without them.

But enough of my meanderings.  This letter may arrive too late for
Christmas, but let me wish you both a Merry Christmas and a Happy New
Year.


\als
\fin


\address
San Francisco Passport Agency--M
525 Market St. \#200
San Francisco, CA 94105
\body
Gentlemen:

I am enclosing an application for a new passport together with my old
passport number K1976320 which expires on June 19, 1984, two recent photographs,
and a check for \$35.

My unseemly haste in asking for a renewal at this early date is caused by
the two months that, it seems, must be allowed to get the necessary visa
for my trip and by the India government's requirement that the visitor
have a passport that does not expire for at least six months after
the intended departure date from their country.

I have just learned of this six-month requirement and am somewhat late in
submitting this application.  I plan to leave the U.S. on February 27th,
and I would appreciate any extra effort that you might make to get my
new passport back to me before December 27th if at all possible.

\als
\fin


\def\\{{\tt \char'134}}
\def\lbr{{\tt \char'173}}  \def\rbr{{\tt \char'175}}  \def\rrbr{{\tt \char'175\char'175}}
\def\h{\hbox{$h$}\llap{\raise.5ex\hbox{--}}}
%\def\h{\hbox{$h$}\hskip-6pt\raise4pt\hbox{--}}
\address
Professor R. N. Bracewell
Durand 329A

\body
Dear Ron:

I happened to come across your short paper on Infrared Planetary Detection
while trying to find something else and I was reminded that I had once
intended to show you a better way to define `\h' (at least in \TeX82), to wit:

\\def\\h\lbr\\hbox\lbr\$h\$\rbr\\llap\lbr\\raise.5ex\\hbox\lbr--\rrbr\rbr
$$\h\omega\Phi={dU\over dx}LF$$

This makes the definition font-size independent and so you would not have to change
the definition if you change the font size.


\als
\fin

\address
Mr. Bevan
The Sequoias

\body
Dear Mr. Bevan:

This letter is prompted by the current concern that many residents
have recently been expressing regarding the decision to locate the pool and Jacuzzi
in the ``down below'' location.

As you know, I have always prefered a ``campus level'' location but I have
gone along with the chosen location because I have felt that a pool
near the croquet court was better than no pool at all.  I have
been maintaining a low profile during recent discussions of this matter,
because I did not want to unduely influence people in their independent
assessment of the merits of the different posible locations.

With time, it has become increasingly clear that there are a substantial
number of our residents who would use a pool located at ``campus level''
but who will not use a pool located ``down below''.  Many of these people
are the very ones who would benefit most by the use of these facilities.

Several residents who fall in this catagory have gone out of their way to
tell me that they would be quite unable to get to and from (and especially
from) the ``down below'' location, that they would very much like to be
able to swim, that they have not contributed to the pool fund because of
its projected location, and, finally that they would make substantial
contributions toward a pool if it were to be at ``campus level''.

Perhaps it is time for us to reassess the situation.  Only two viable
``campus level'' sites exist. The one near Building 17 is probably too unpopular
with some of the nearby residents, but the present site of the cutting
garden has never been given serious consideration.

Factors to be considered are: 

1) The cutting garden would have to be relocated or abandoned.  Betty Hone
has stated that she will not manage this after April 1st and that she has been
unable to find a successor.  If necessary, the cutting garden could be
relocated in a number of places, the area to the north west of Building 1
would be perhaps the best.

2) The nearness to residential areas would allow for some reduction in the
elaborateness of the dressing rooms and toilet facilities,
with a substantial saving in costs.

3) The cutting garden area is a bit on the small size, but, in view of 2) above,
the area is still adequate.

4) The pool structure would not be visible from outside of our property, and
the dangers of vandalism would be considerably reduced. Surveillance by our
night watchman would be easier.

5) We would have very much less trouble in getting town of Portola Valley
approval for the project.

6) There is a lack of a usable parking area near the cutting garden
location.  This objection has no merit.  Residents who still own cars
usually have to walk substantial distances to get their cars, and many of
the people who would need tranportation, no longer have cars of
their own.  Parking might be a consideration ``down below'' but it is not at
campus level.

7) While the distance from the health center is somewhat greater that for
the ``down below'' location, residents in the health center would have to
be provided with transportation for either location.

8) Electrical service, water, sewer, and storm-drain connections pose no
particular problems.  Some costs would be involved in providing a gas
connection.

9) There would be no problem in finding locations for solar pamels, either
on the nearby car-port roofs or within the pool structure itself, in
accordance with my recent suggestion which would still be usable in this location.

10) The greater use that would be made of the pool and Jacuzzi would help to
maintain the health and good spirits of our residents, and incidentally, give us a
larger base over which to distribute the maintenance costs.

In view of all the considerations that I have outlined, I would like to recommend
that the pool location be reconsidered. Perhaps the architect should be consulted
in this regard and be given an opportunity to inspect this alternate site.

\als

\annotations
c/c Members of the Pool Committee
\fin


\address
Mr. J. A. Goetz
202D The Commons Way
Fishkill, NY 12524

\body
Dear Jack:

Merry Christmas to you and to Dorothy.

I was away on a trip when your letter of August 24th arrived and I did not
get back in time to answer it before you could have been on your way to
the west coast. I did expected to hear from you while you were in the Palo
Alto area.  I am sorry to be so slow in answering, but I have been away a
lot and extra busy while at home.

I am pleased to learn that the Mid-Hudson Section is doing so well.  At
the moment, I am at loss for anything to offer by way of a recollection of
the early days of the Section.  Somehow or other, I seem to have involved
myself in starting a number of different enterprises that others have made
prosper and I seem to get more credit than I deserve when it is the people
who hang on and make the enterprise a success, such as youself, who deserve
the credit.

I was glad to hear that you are both hale and hearty.  

Bernice died in May of 1982.  She had a heart bypass operation in January
of that year and she never fully recovered. I miss her greatly.  Fortunately,
we moved into a retirement community some years ago, so I am not as much alone
as I would be if I were living in a single family home.

We had planned to go around the world on the Rotterdam in 1982.  When
Bernice first took sick, we postponed the trip a year and after Bernice's
death, Margaret and Donna (our two daughters, do you remember them)
insisted that I simply must take the trip alone.

So I started this year off with a long trip. I took an IBM-pc with me
and I gave a series of lectures and demonstrations on the ship.  I had taught a
class on the Rotterdam in 1971 (that time on the game of Go), so this was
not a particularly new thing for me to do.  It did keep me busy and it was
a good way to get acquainted with the passengers, something that might
otherwise been hard for me to do.

At 82, I am still working, and I have now been at Stanford a little longer than I
was at IBM, 17 years at IBM and 17 years and a few months at Stanford.
I have been unusually lucky in being able to get a job here at Stanford
after I retired and to hang on to it for these many years.

In a sense, I suppose, I have retired, in that I am very much my own boss
and only work when I want to.  At my own insistance, I have been
gradually reducing my work week so that I am now officially working only
one fourth time although I still put in a 40 or 50 hour week when I am at
home.

I do take a lot of time off for trips, having traveled over 45,000 miles
so far this year.  Being away as much as I am, I have given up teaching
and I spend my time doing odd research jobs and just being a sort of handy
man around the department.  For example, I have just finished writing a
beginners manual explaining the typesetting system, called \TeX, on which
I have done a small amount of work.

This letter will be formatted with \TeX, although it will not be a very
good example as I will print it on an on-line printer that we use
for run-of-the-mill work.  This printer prints 384 dots to the inch.  We use a
more elaborate printer at 1000 dots to the inch, or one
at 5333 dots to the inch when producing masters for books that will be
reproduced by photo-offset.
\TeX\ is particularly good for text that involves
mathematical equations. It allows one to use up to 256 different fonts
each with 256 characters, on any one document, and we have well over 500
different fonts available.

Let me again wish you a Merry Christmas.

\als

\annotations

\ps
p.s. I cannot refrain from enclosing two pages from the beginners \TeX\ manual
to show some of the special mathematical symbols that
are available.

\fin


\personal 
%\stanford

\address
The Bank of New York, Trustee
Municipal Investment Trust Fund, 1st Calif.
Att. Mr. Carlos Maldonado
Corporate Trust Division
P.O. Box 11,002
Church Street Station
New York, N.Y. 10249

\body
Dear Sir:

As requested in your letter of 11/28/83, I am returning herewith the
documents that you named, these being, the
Certificate of Ownership, Certificate 1 CA 02442 dated Nov. 1, 1983, for 10
units, Mutual Investment Trust Fund, First California Series, account
number 268 10669930-9, held in the names of Arthur L.  Samuel and Bernice
D. Samuel JT TEN., an Affidavit of Domicile and a certified
copy of the death certificate for my wife Bernice D. Samuel who died on
May 24, 1982.

Will you please transfer owernership of the ten units covered by this
certificate to my name, Arthur L. Samuel residing at 501 Portola Road,
Portola Vally, CA 94025, social security number 091-09-8617 and mail the
new certificate of ownership and the monthly checks to me at my mail
address, namely 501 Portola Road, Box 8214, Portola Valley, CA 94025.

Mrs. Samuel died without leaving a valid will. All claims againt the
estate, funeral and administration expenses have been paid.  There were no
Inheritance and Estate Taxes charged against the estate and by court order
I am listed as her sole heir.

\als
\fin

\address
Mr. Sanjoy Mahajan
121 Wilmar Drive
Pittsburgh, PA 15238
\body
Dear Mr. Mahajan:

This is in reply to your letter of November 21, 1983 in regard to my
checker program.  It has been many years since I last worked on this
program. While there have been numerous attempts made to improve the
program, both by me and by others, the set of parameters as described in
these earlier papers remained fairly good.  As you have probably noticed,
my primary purpose was to study the learning process and the goodness of
play was always a secondary consideration.

About the only thing that I have done since my 1967 paper has been to work
with a few master checker players in an attempt to extract from them the
thought processes that lead them to arrive at their moves.  I have been
singularly unsuccessful in this endeavor. Master checker players seem to
know intuitively what is the best move and when they attempt to
explain the reasons for their moves they usually dredge up a line of
reasoning that seems to have little or nothing to do with the actual
thought processes that lead to their choice.

I am somewhat at loss as to how to answer your specific questions.  Many
of them relate to matters that are covered in detail in the two papers
that you have read and I would be only repeating what I wrote.  Since my
work was related to learning, I actually tried a great many different
schemes for making the calculations of (1) piece advantage, (2) ways of
defining terms and choice of terms, (3) setting value ranges and actual
values as used in the signature tables, (4) setting cut-off points, (5)
the number of different signature tables to use and the decision as to
which one to use, etc.. I could go on referencing your listed questions
but perhaps this is enough to indicate how really impossible it is for me
to answer your questions in any simple fashion.  

I would be very glad to give you a copy of my checker
program if I thought that
you could use it. Unfortunately, it is written in an assembly language for
the DEC 10 and it contains over 40,000 poorly-commented instructions.
It really would only confuse you. Through the years, I have made it
available to many different people and the only people who have been able to
use it were those who had DEC 10's and who were familiar
with its inner workings and with its machine code.
Incidentally, Slagle, whose work on alpha-beta
pruning you mentioned in (9), was one of these, and I was well acquainted
with his work.

I am not sure what you mean when you say that you would like to implement
my program.  If you simply want to write a program that will play a
reasonably good game of checkers, then the linear-polynomial scheme
works fairly well.  If, on the other hand, you want to study machine learning
then you have a long and complicated program ahead of you and you should
review all of the work in AI before settling on a game as your study
vehicle. I still favor games but there are others who do not agree with me.

It should be possible to write a fairly good IBM-PC checker program, I
have a PC at home and I have occasionally thought about doing this.  As
evidence of what can be done, I once wrote a simple checker program
(without learning) for an early home game computer (called the VideoBrain)
that is no longer on the market. The entire program was tailored to fit in
8k bytes of rom memory. This program uses a joy stick for input, it
displays the board on a TV, and it plays a surprisingly good game.
Unfortunately, I am not at liberty to supply copies of this program but it
would be of little help in any case, being written in the assembly code for
a specific chip, and for compilation by an unavailable cross compiler on
an unavailable computer.

Were I to start working on checkers again, I would write the
program in some higher language, probably PASCAL or perhaps C or ADA, to
simplify the programming task and to insure transportability, and accept the
degradation in playing speed that would result.  I might try PROLOG simply
to learn it characteristics.
LISP is often the favored language for
AI research but I would not recommend it for checkers unless one had
access to one of the newer LISP machines.

You did not tell me the make ofcomputer that you intend to buy.  I have an
IBM-PC at home and I have occasionally thought about writing a checker
program for it. As evidence of what can be done, I once wrote a simple
checker program (without learning) for an early home game computer (called
the VideoBrain) that is no longer on the market. The entire program is
tailored to fit in 8k bytes of rom memory and it uses a joy stick for
input and displays the board on a TV. It plays a surprisingly good game
although not good enough to interest you.

\als

\annotations

\ps

\fin

\def\today{\ifcase\month\or
  January\or February\or March\or April\or May\or June\or
  July\or August\or September\or October\or November\or December\fi
  \space\number\day, \number\year}

\raggedbottom
\interlinepenalty=1000
\hsize=6.25truein
\voffset=24pt
\advance\vsize by-\voffset
\parindent=0pt
\parskip=0pt
\nopagenumbers
\headline={\ifnum\pageno>1
  \tenrm To \addressee\hfil\today\hfil Page \folio
  \else\hfil\fi}

\def\beginlinemode{\endmode
  \begingroup\obeylines\def\endmode{\par\endgroup}}
\def\beginparmode{\endmode
  \begingroup\parskip=\medskipamount \def\endmode{\par\endgroup}}
\let\endmode=\par
\def\endletter{\endmode\vfill\supereject}

\newdimen\longindentation \longindentation=4truein
\newbox\theaddress
\def\address{\beginlinemode\getaddress}
{\obeylines\gdef\getaddress #1
  #2
  {#1\gdef\addressee{#2}%
    \global\setbox\theaddress=\vbox\bgroup\raggedright%
    \hsize=\longindentation \everypar{\hangindent2em}#2
    \def\endmode{\egroup\endgroup \copy\theaddress \bigskip}}}

\def\body{\beginparmode}
\def\closing{\beginlinemode\getclosing}
{\obeylines\gdef\getclosing #1
  #2
  {#1\nobreak\bigskip \leftskip=\longindentation #2
    \nobreak\bigskip\bigskip\bigskip % space for signature
    \def
    {\endgraf\nobreak}}}
\def\annotations{\beginlinemode\def\par{\endgraf\nobreak}\obeylines\par}
\def\ps{\beginparmode\nobreak
  \interlinepenalty5000\def\par{\endgraf\penalty5000}}

\def\up#1{\leavevmode \raise.16ex\hbox{#1}}
\font\smallheadfont=cmr8 at 8truept
\font\largeheadfont=cmdunh10 at 14.4truept
\font\logofont=manfnt at 14.4truept

\def\alsletterhead{
  \def\sendingaddress{Arthur L. Samuel\par
    The Sequoias\par
    501 Portola Road, box 8214\par
    Portola Valley, CA 94025\par
    \up[415\up]\thinspace 424-4233\par}
  \def\returnaddress{A. L. Samuel\par
    The Sequoias\par
    501 Portola Road, box 8214\par
    Portola Valley, CA 94025 USA}
  \letterhead}

\def\letterhead{\pageno=1 \def\addressee{} \univletterhead
  {\leftskip=\longindentation
    {\baselineskip9truept\smallheadfont\sendingaddress}
    \bigskip\bigskip\rm\today\bigskip}}

\def\univletterhead{\vglue-\voffset
\font\seal=stan70
  \hbox{\hbox to\longindentation{\raise4truemm\hbox{\seal
        \kern2truept X\kern-1.667truept
        \lower2truept\hbox{X}\kern-1.667truept X}\hfil
      \largeheadfont Stanford University\hfil}%
    \kern-\longindentation
    \vbox{\smallheadfont\baselineskip9truept
      \leftskip=\longindentation STANFORD, CA 94305}}
  \vskip2truept\hrule\vskip4truept }

\def\makelabel{\endletter\hbox{\vrule
    \vbox{\hrule \kern6truept
      \hbox{\kern6truept\vbox to 2truein{\hsize=\longindentation
          \smallheadfont\baselineskip9truept\returnaddress
          \vfill\moveright 2truein\copy\theaddress\vfill}%
        \kern6truept}\kern6truept\hrule}\vrule}
  \pageno=0\vfill\eject}

Let's consider {business letters} first. Suppose that you want \TeX\ to
format your {correspondence}, and that you have $n$~letters to send. {mail}
If your computer system contains a file letterformat.tex like the one
described later in this appendix, it's easy to do the job by applying \TeX\ to
a file that looks like this:
\begindisplay
\<optional magnification>\cr
\input letterformat\cr
\<business letter$_1$>\cr
\noalign{\vskip-1pt}
\qquad\vdots\cr
\<business letter$_n$>\cr
\end\cr
\enddisplay
Here each of the $n$ business letters has the form
\begindisplay
\<letterhead>\cr
\address\cr
\<one or more lines of address>\cr
\body\cr
\<one or more paragraphs of text>\cr
\closing\cr
\<one or more lines for salutation and signature>\cr
\<optional annotations>\cr
\<optional postscripts>\cr
\endletter\cr
\makelabel % omit this if you don't want an address label\cr
\enddisplay
The \<letterhead> at the beginning of this construction is usually a
control sequence like \rjdletterhead for letters by R.~J.~D.; each
letter writer can have a personalized letterhead that is stored with the
letterformat macros. The \<optional annotations> at the end are any number
of one-line notes preceded by `\annotations'; the \<optional postscripts>
are any number of paragraphs preceded by `\ps'. When \TeX\ is processing the
\address and the \closing and the optional \annotations, it produces
output line-for-line just as the lines appear in the input file; but when
\TeX\ is processing the \body of the letter and the optional \ps, it
chooses line breaks and justifies lines as it normally does when
typesetting paragraphs in books.

A complete example, together with the resulting output, appears on the
next two pages. This example starts with `\magnification=\magstep1'
because the letter is rather short. Magnification is usually omitted if the
letters are long-winded; `\magnification=\magstephalf' is
appropriate when they are medium-size. The same magnification applies to
all~$n$ letters, so you must run \TeX\ more than once if you want more than
one magnification.

\eject
\begingroup \obeylines

\magnification=\magstep1
\input letterformat
\bigskip
\rjdletterhead % (see the output on the next page)
\bigskip
\address
Prof.~Brian~K. Reid{Reid}
Department of Electrical Engineering
Stanford University
Stanford, CA 94305
\bigskip
\body
Dear Prof.~Reid:
\bigskip
I understand that you are having difficulties with
Alka-Seltzer tablets.  Since there are 25~pills{Alka-Seltzer}{ties}
per bottle, while the manufacturer's directions
recommend ``plop,~plop, fizz,~fizz,'' my colleagues
tell me that you have accumulated a substantial
number of bottles in which there is one tablet
left. % (See the 1978 SCRIBE User Manual, page 90.){Scribe}
\bigskip
At present I am engaged in research on the potential
applications of isolated analgesics. If you would
be so kind as to donate your Alka-Seltzer collection
to our project, I would be more than happy to send
you preprints of any progress reports that we may
publish concerning this critical problem.
\bigskip
\closing
Sincerely,
R. J. Drofnats{Drofnats}
Professor

\annotations
RJD/dek
cc: {\sl The \TeX book}

\ps
P. S. \ If you like, I will check into the
possibility that your donation and the meals that
you have been eating might be tax-deductible, in{IRS}
connection with our research.
\endletter
\makelabel
\eject\endgroup

\def\proofcopy(#1){\ifproofmode\raise.5in\hbox{\sevenrm(#1)}\else\hfil\fi}
\vglue-\topskip \nointerlineskip
\dimen0=\vsize \advance\dimen0 by-1.2pt \advance\dimen0 by-2.1in
\vbox{\hrule
  \hbox{\vrule height \dimen0
    \hbox to 4.25in{\hss\proofcopy(Output page goes here, reduced 50\%.)\hss}
    \vrule}}
\nointerlineskip
\hrule
\line{\vrule height 2.1in
  \hss\proofcopy(Label and stamp go here, reduced 50\%.)\hss\vrule}
\hrule
\vskip 0pt plus .001pt minus .001pt % in case of rounding errors
\eject
If the letter is more than one page long, the addressee, date, and
page number will appear at the top of subsequent pages. For example,
the previous letter comes out as follows, if additional paragraphs are
added to the text:
\medskip
\hrule
\line{\vrule height 3.1127in
  \hfil\proofcopy(First page, reduced to 28.3\%.)\hfil\vrule\hfil
  \proofcopy(Second page, reduced to 28.3\%.)\hfil\vrule}
\hrule
\bigskip
\ninepoint
The macro package letterformat.tex that produces this format begins
with a simple macro that expands to the current {date}.
\beginlines
\def\today{\ifcase\month\or
  January\or February\or March\or April\or May\or June\or
  July\or August\or September\or October\or November\or December\fi
  \space\number\day, \number\year}
\endlines

Then comes the specification of page layout, which is ``ragged'' at the
bottom. A rather large \interlinepenalty is used so that page
breaks will tend to occur between paragraphs.
\beginlines
\raggedbottom
\interlinepenalty=1000
\hsize=6.25truein
\voffset=24pt
\advance\vsize by-\voffset
\parindent=0pt
\parskip=0pt
\nopagenumbers
\headline={\ifnum\pageno>1
  \tenrm To \addressee\hfil\today\hfil Page \folio
  \else\hfil\fi}
\endlines

The contents of a letter are typeset either in ``line mode'' (obeying lines)
or in ``paragraph mode'' (producing paragraphs in {block style}). Control
sequences \beginlinemode and \beginparmode are defined to initiate these
modes; and other control sequence, \endmode, is defined and redefined so
that the current mode will terminate properly:
\beginlines
\def\beginlinemode{\endmode
  \begingroup\obeylines\def\endmode{\par\endgroup}}
\def\beginparmode{\endmode
  \begingroup\parskip=\medskipamount \def\endmode{\par\endgroup}}
\let\endmode=\par
\def\endletter{\endmode\vfill\supereject}
\endlines

One of the chief characteristics of this particular business letter format
is a parameter called \longindentation, which is used to indent the
closing material, the date, and certain aspects of the letterhead. The
\address macro creates a box that will be used both in the letter and in
the label on the envelope. If individual lines of the address exceed
\longindentation, they are broken, and hanging indentation is used for any
material that must be carried over.
\beginlines
\newdimen\longindentation \longindentation=4truein
\newbox\theaddress
\def\address{\beginlinemode\getaddress}
{\obeylines\gdef\getaddress #1
  #2
  {#1\gdef\addressee{#2}%
    \global\setbox\theaddress=\vbox\bgroup\raggedright%
    \hsize=\longindentation \everypar{\hangindent2em}#2
    \def\endmode{\egroup\endgroup \copy\theaddress \bigskip}}}
\endlines
(Parameter #2 to \getaddress {parameters, delimited}
will be the contents of the line following \address, i.e., the
name of the addressee.)

The closing macros are careful not to allow a page break anywhere between the
end of the \body and the beginning of a \ps.
\beginlines
\def\body{\beginparmode}
\def\closing{\beginlinemode\getclosing}
{\obeylines\gdef\getclosing #1
  #2
  {#1\nobreak\bigskip \leftskip=\longindentation #2
    \nobreak\bigskip\bigskip\bigskip % space for signature
    \def
    {\endgraf\nobreak}}}
\def\annotations{\beginlinemode\def\par{\endgraf\nobreak}\obeylines\par}
\def\ps{\beginparmode\nobreak
  \interlinepenalty5000\def\par{\endgraf\penalty5000}}
\endlines

The remaining portion of letterformat.tex deals with {letterheads} and
labels, which of course will be different for different organizations.
The following macros were used to generate the examples in this
appendix; they can be modified in more-or-less obvious ways to produce
suitable letterheads of other kinds. Special fonts are generally
needed, and they should be loaded at `true' sizes so that they are not
affected by magnification.  One tiny refinement worth noting here is the
\up macro, which raises {brackets} so that they look better in a
{telephone number}.
\beginlines
\def\up#1{\leavevmode \raise.16ex\hbox{#1}}
\font\smallheadfont=cmr8 at 8truept
\font\largeheadfont=cmdunh10 at 14.4truept
\font\logofont=manfnt at 14.4truept
\smallbreak
\def\rjdletterhead{
  \def\sendingaddress{R. J. DROFNATS, F.T.U.G.\par{TeX Users Group}
    PROFESSOR OF FARM ECOLOGY\par
    TEX.RJD @ SU-SCORE.ARPA\par{atsign}
    \up[415\up]\thinspace 497-4975\par}
  \def\returnaddress{R. J. Drofnats, Dept.~of Farm Ecology\par
    The University of St.~Anford\par
    P. O. Box 1009, Haga Alto, CA 94305 USA}
  \letterhead}
\smallbreak
\def\letterhead{\pageno=1 \def\addressee{} \univletterhead
  {\leftskip=\longindentation
    {\baselineskip9truept\smallheadfont\sendingaddress}
    \bigskip\bigskip\rm\today\bigskip}}
\smallbreak
\def\univletterhead{\vglue-\voffset
  \hbox{\hbox to\longindentation{\raise4truemm\hbox{\logofont
        \kern2truept X\kern-1.667truept
        \lower2truept\hbox{X}\kern-1.667truept X}\hfil
      \largeheadfont The University of St.~Anford\hfil}%
    \kern-\longindentation
    \vbox{\smallheadfont\baselineskip9truept
      \leftskip=\longindentation BOX 1009\par HAGA ALTO, CA 94321}}
  \vskip2truept\hrule\vskip4truept }
\smallbreak
\def\makelabel{\endletter\hbox{\vrule
    \vbox{\hrule \kern6truept
      \hbox{\kern6truept\vbox to 2truein{\hsize=\longindentation
          \smallheadfont\baselineskip9truept\returnaddress
          \vfill\moveright 2truein\copy\theaddress\vfill}%
        \kern6truept}\kern6truept\hrule}\vrule}
  \pageno=0\vfill\eject}
\endlines

\beginchapter Appendix E. Example Formats

Although the plain \TeX\ format of Appendix B is oriented to technical
reports, it can readily be adapted to quite different applications. Examples of
three such adaptations are provided in this appendix: (1)~a~format for
business letters; (2)~a~format for concert programs; (3)~the~format used
to typeset this book.

Let's consider {business letters} first. Suppose that you want \TeX\ to
format your {correspondence}, and that you have $n$~letters to send. {mail}
If your computer system contains a file letterformat.tex like the one
described later in this appendix, it's easy to do the job by applying \TeX\ to
a file that looks like this:
\begindisplay
\<optional magnification>\cr
\input letterformat\cr
\<business letter$_1$>\cr
\noalign{\vskip-1pt}
\qquad\vdots\cr
\<business letter$_n$>\cr
\end\cr
\enddisplay
Here each of the $n$ business letters has the form
\begindisplay
\<letterhead>\cr
\address\cr
\<one or more lines of address>\cr
\body\cr
\<one or more paragraphs of text>\cr
\closing\cr
\<one or more lines for salutation and signature>\cr
\<optional annotations>\cr
\<optional postscripts>\cr
\endletter\cr
\makelabel % omit this if you don't want an address label\cr
\enddisplay
The \<letterhead> at the beginning of this construction is usually a
control sequence like \rjdletterhead for letters by R.~J.~D.; each
letter writer can have a personalized letterhead that is stored with the
letterformat macros. The \<optional annotations> at the end are any number
of one-line notes preceded by `\annotations'; the \<optional postscripts>
are any number of paragraphs preceded by `\ps'. When \TeX\ is processing the
\address and the \closing and the optional \annotations, it produces
output line-for-line just as the lines appear in the input file; but when
\TeX\ is processing the \body of the letter and the optional \ps, it
chooses line breaks and justifies lines as it normally does when
typesetting paragraphs in books.

A complete example, together with the resulting output, appears on the
next two pages. This example starts with `\magnification=\magstep1'
because the letter is rather short. Magnification is usually omitted if the
letters are long-winded; `\magnification=\magstephalf' is
appropriate when they are medium-size. The same magnification applies to
all~$n$ letters, so you must run \TeX\ more than once if you want more than
one magnification.

\eject
\begingroup \obeylines
\magnification=\magstep1
\input letterformat
\bigskip
\rjdletterhead % (see the output on the next page)
\bigskip
\address
Prof.~Brian~K. Reid{Reid}
Department of Electrical Engineering
Stanford University
Stanford, CA 94305
\bigskip
\body
Dear Prof.~Reid:
\bigskip
I understand that you are having difficulties with
Alka-Seltzer tablets.  Since there are 25~pills{Alka-Seltzer}{ties}
per bottle, while the manufacturer's directions
recommend ``plop,~plop, fizz,~fizz,'' my colleagues
tell me that you have accumulated a substantial
number of bottles in which there is one tablet
left. % (See the 1978 SCRIBE User Manual, page 90.){Scribe}
\bigskip
At present I am engaged in research on the potential
applications of isolated analgesics. If you would
be so kind as to donate your Alka-Seltzer collection
to our project, I would be more than happy to send
you preprints of any progress reports that we may
publish concerning this critical problem.
\bigskip
\closing
Sincerely,
R. J. Drofnats{Drofnats}
Professor
\bigskip
\annotations
RJD/dek
cc: {\sl The \TeX book}
\bigskip
\ps
P. S. \ If you like, I will check into the
possibility that your donation and the meals that
you have been eating might be tax-deductible, in{IRS}
connection with our research.
\endletter
\makelabel
\eject\endgroup
\def\proofcopy(#1){\ifproofmode\raise.5in\hbox{\sevenrm(#1)}\else\hfil\fi}
\vglue-\topskip \nointerlineskip
\dimen0=\vsize \advance\dimen0 by-1.2pt \advance\dimen0 by-2.1in
\vbox{\hrule
  \hbox{\vrule height \dimen0
    \hbox to 4.25in{\hss\proofcopy(Output page goes here, reduced 50\%.)\hss}
    \vrule}}
\nointerlineskip
\hrule
\line{\vrule height 2.1in
  \hss\proofcopy(Label and stamp go here, reduced 50\%.)\hss\vrule}
\hrule
\vskip 0pt plus .001pt minus .001pt % in case of rounding errors
\eject
If the letter is more than one page long, the addressee, date, and
page number will appear at the top of subsequent pages. For example,
the previous letter comes out as follows, if additional paragraphs are
added to the text:
\medskip
\hrule
\line{\vrule height 3.1127in
  \hfil\proofcopy(First page, reduced to 28.3\%.)\hfil\vrule\hfil
  \proofcopy(Second page, reduced to 28.3\%.)\hfil\vrule}
\hrule
\bigskip
\ninepoint
The macro package letterformat.tex that produces this format begins
with a simple macro that expands to the current {date}.
\beginlines
\def\today{\ifcase\month\or
  January\or February\or March\or April\or May\or June\or
  July\or August\or September\or October\or November\or December\fi
  \space\number\day, \number\year}
\endlines

Then comes the specification of page layout, which is ``ragged'' at the
bottom. A rather large \interlinepenalty is used so that page
breaks will tend to occur between paragraphs.
\beginlines
\raggedbottom
\interlinepenalty=1000
\hsize=6.25truein
\voffset=24pt
\advance\vsize by-\voffset
\parindent=0pt
\parskip=0pt
\nopagenumbers
\headline={\ifnum\pageno>1
  \tenrm To \addressee\hfil\today\hfil Page \folio
  \else\hfil\fi}
\endlines

The contents of a letter are typeset either in ``line mode'' (obeying lines)
or in ``paragraph mode'' (producing paragraphs in {block style}). Control
sequences \beginlinemode and \beginparmode are defined to initiate these
modes; and other control sequence, \endmode, is defined and redefined so
that the current mode will terminate properly:
\beginlines
\def\beginlinemode{\endmode
  \begingroup\obeylines\def\endmode{\par\endgroup}}
\def\beginparmode{\endmode
  \begingroup\parskip=\medskipamount \def\endmode{\par\endgroup}}
\let\endmode=\par
\def\endletter{\endmode\vfill\supereject}
\endlines

One of the chief characteristics of this particular business letter format
is a parameter called \longindentation, which is used to indent the
closing material, the date, and certain aspects of the letterhead. The
\address macro creates a box that will be used both in the letter and in
the label on the envelope. If individual lines of the address exceed
\longindentation, they are broken, and hanging indentation is used for any
material that must be carried over.
\beginlines
\newdimen\longindentation \longindentation=4truein
\newbox\theaddress
\def\address{\beginlinemode\getaddress}
{\obeylines\gdef\getaddress #1
  #2
  {#1\gdef\addressee{#2}%
    \global\setbox\theaddress=\vbox\bgroup\raggedright%
    \hsize=\longindentation \everypar{\hangindent2em}#2
    \def\endmode{\egroup\endgroup \copy\theaddress \bigskip}}}
\endlines
(Parameter #2 to \getaddress {parameters, delimited}
will be the contents of the line following \address, i.e., the
name of the addressee.)

The closing macros are careful not to allow a page break anywhere between the
end of the \body and the beginning of a \ps.
\beginlines
\def\body{\beginparmode}
\def\closing{\beginlinemode\getclosing}
{\obeylines\gdef\getclosing #1
  #2
  {#1\nobreak\bigskip \leftskip=\longindentation #2
    \nobreak\bigskip\bigskip\bigskip % space for signature
    \def
    {\endgraf\nobreak}}}
\def\annotations{\beginlinemode\def\par{\endgraf\nobreak}\obeylines\par}
\def\ps{\beginparmode\nobreak
  \interlinepenalty5000\def\par{\endgraf\penalty5000}}
\endlines

The remaining portion of letterformat.tex deals with {letterheads} and
labels, which of course will be different for different organizations.
The following macros were used to generate the examples in this
appendix; they can be modified in more-or-less obvious ways to produce
suitable letterheads of other kinds. Special fonts are generally
needed, and they should be loaded at `true' sizes so that they are not
affected by magnification.  One tiny refinement worth noting here is the
\up macro, which raises {brackets} so that they look better in a
{telephone number}.
\beginlines
\def\up#1{\leavevmode \raise.16ex\hbox{#1}}
\font\smallheadfont=cmr8 at 8truept
\font\largeheadfont=cmdunh10 at 14.4truept
\font\logofont=manfnt at 14.4truept
\smallbreak
\def\rjdletterhead{
  \def\sendingaddress{R. J. DROFNATS, F.T.U.G.\par{TeX Users Group}
    PROFESSOR OF FARM ECOLOGY\par
    TEX.RJD @ SU-SCORE.ARPA\par{atsign}
    \up[415\up]\thinspace 497-4975\par}
  \def\returnaddress{R. J. Drofnats, Dept.~of Farm Ecology\par
    The University of St.~Anford\par
    P. O. Box 1009, Haga Alto, CA 94305 USA}
  \letterhead}
\smallbreak
\def\letterhead{\pageno=1 \def\addressee{} \univletterhead
  {\leftskip=\longindentation
    {\baselineskip9truept\smallheadfont\sendingaddress}
    \bigskip\bigskip\rm\today\bigskip}}
\smallbreak
\def\univletterhead{\vglue-\voffset
  \hbox{\hbox to\longindentation{\raise4truemm\hbox{\logofont
        \kern2truept X\kern-1.667truept
        \lower2truept\hbox{X}\kern-1.667truept X}\hfil
      \largeheadfont The University of St.~Anford\hfil}%
    \kern-\longindentation
    \vbox{\smallheadfont\baselineskip9truept
      \leftskip=\longindentation BOX 1009\par HAGA ALTO, CA 94321}}
  \vskip2truept\hrule\vskip4truept }
\smallbreak
\def\makelabel{\endletter\hbox{\vrule
    \vbox{\hrule \kern6truept
      \hbox{\kern6truept\vbox to 2truein{\hsize=\longindentation
          \smallheadfont\baselineskip9truept\returnaddress
          \vfill\moveright 2truein\copy\theaddress\vfill}%
        \kern6truept}\kern6truept\hrule}\vrule}
  \pageno=0\vfill\eject}
\endlines