card-lab.tex includes two diagrams produced using Diagram!.  The files
selection-sort.diagram and merge.diagram are the source form of these
diagrams, but won't be useful to others without this software.  The
corresponding *.eps files are what are actually included in the
card-lab.tex document.  Even to TeX and print that you will need dvips
and the epsf.sty document style option that comes with it.  For those
with none of the above, the card-lab.ps is a pre dvips'ed version,
ready to print on any 300dpi postscript printer.

The other .diagram file, cards.diagram, isn't part of the document at
all, but was used to produce the special deck of cards that we used
with this lab.  There is again a ready-to-print PostScript version,
cards.ps.

For those wanting to make their own decks of cards form scratch:
 - We put 12 2.125 x 3.5 inch cards on each 8.5x11 sheet of card stock, with
  .5 x 8.5 inches wasted on each sheet; this seems to be a good size.
 - We did 36 cards (3 sheets) in each deck; this allows some variety in
   the cards the students actually wind up sorting if they shuffle first
   and then count of 16, e.g.
 - The 36 numbers were randomly chosen among those from 0-99, but with
   the fudging that there were no duplicates (which complicates matters)
   and no two consecutive numbers (which allows students to know that a
   card must be the next one without looking at the rest).
 - We took care to underline any numbers that could be misinterpreted if
   held upside-down (e.g. 6 vs. 9).

The file graph.scm was used by the lab instructor to graphically show
the students the results.  This file is specific to the Schematik
scheme environment on the NeXT, but could be easily ported to MIT Scheme
on other systems or with more effort to another Scheme with graphics.
It was hacked up quickly the night before the lab and is purely utilitarian,
not exemplary at all.

What we did was keep a running list of all the times reported by the
various teams, rolling along from one lab section to the next.  That
way even the first team in a lab section could immediately upon
reporting their data see it in the context of other teams' data.  The
graph shows the entire range of times reported by the various teams as
a vertical bar, with a horizontal cross tick at the average.  The two
sorting algorithms are shown in two different colors (green for
mergesort, red for selection sort; or shades of gray on a monochrome
system) and are displaced slightly in opposite directions from the true
x position to prevent overlap.

In addition to discussing this graph with the students, we asked the
question: "How would you feel if I asked you to selection sort 32
cards now?"  This generally made them recognize their gut feelings
that it was growing worse fast as the size increased.  We then in
class the next day talked more formally about analyzing the two
algorithms.
