The following results from linear discriminant analysis (run in R) can
 be found at the bottom of this file and are summarized as follows:
 
 [1] m m m m m m m m m m m m

i.e. all 12 disputed papers are classified into the "madison" class

Note that the four features tested are the top 2 function words
("upon" and "there") and the top 2 word length features (2-letter
words and 1-letter words).  These rankings were computed
using contigency tables.
*****************************************************************


> test <- read.table("madham.txt", sep="");
> train <- read.table("disputed.txt", sep="");
> cl <- factor(c(rep("m",14), rep("h",18)))
> test
      V1    V2     V3      V4
1  0.000 2.003 27.036 206.943
2  0.000 0.000 18.250 212.915
3  0.369 0.738 21.033 202.583
4  1.207 0.905 29.261 211.161
5  0.000 0.000 20.353 236.943
6  0.000 1.328 18.254 216.064
7  0.000 0.282 26.494 204.622
8  0.720 1.081 18.732 211.095
9  0.000 0.583 30.877 218.177
10 0.000 1.379 18.276 203.793
11 0.000 1.418 10.875 201.891
12 0.000 0.000 17.625 200.383
13 0.000 1.097 14.991 209.506
14 0.000 1.077 23.694 210.016
15 3.788 1.263 24.621 224.116
16 1.876 3.752 26.266 189.962
17 4.885 3.996 20.870 215.364
18 1.511 1.007 23.162 212.991
19 2.026 1.520 26.849 219.352
20 2.406 3.208 28.869 212.911
21 3.265 4.198 22.388 215.019
22 3.853 3.302 27.518 238.305
23 2.523 3.364 23.129 241.800
24 2.377 6.452 20.034 232.937
25 1.580 3.686 27.383 232.227
26 4.965 2.483 22.344 231.380
27 4.955 2.252 24.324 232.432
28 2.127 4.558 24.309 216.651
29 1.764 1.764 27.043 230.453
30 2.458 4.425 28.024 226.647
31 3.015 4.020 30.151 235.176
32 5.068 1.521 36.999 231.627
> train
      V1    V2     V3      V4
1  0.000 1.217 21.290 226.277
2  0.908 0.000 17.257 205.268
3  0.000 2.093 25.118 225.536
4  0.000 0.000 21.727 217.273
5  0.000 0.926 23.611 208.796
6  1.003 0.501 19.549 232.581
7  0.000 2.458 29.007 210.914
8  0.000 2.454 29.448 192.638
9  0.000 1.816 18.611 226.509
10 0.000 0.960 28.325 207.393
11 0.000 0.000 34.439 220.076
12 0.000 2.639 22.757 213.391
> cl
 [1] m m m m m m m m m m m m m m h h h h h h h h h h h h h h h h h h
Levels: h m
> z <- lda(test, cl)
Error: couldn't find function "lda"
> library(MASS)
> z <- lda(test, cl)
> z
Call:
lda.data.frame(test, cl)

Prior probabilities of groups:
     h      m 
0.5625 0.4375 

Group means:
        V1        V2       V3       V4
h 3.024556 3.1539444 25.79350 224.4083
m 0.164000 0.8493571 21.12507 210.4351

Coefficients of linear discriminants:
           LD1
V1 -0.78164820
V2 -0.58883622
V3 -0.06020593
V4 -0.01141801
> predict(z, train)
$class
 [1] m m m m m m m m m m m m
Levels: h m

$posterior
              h         m
1  1.160428e-03 0.9988396
2  1.612292e-04 0.9998388
3  2.227611e-02 0.9777239
4  4.739776e-05 0.9999526
5  4.569786e-04 0.9995430
6  4.370276e-03 0.9956297
7  6.637281e-02 0.9336272
8  3.267559e-02 0.9673244
9  2.535042e-03 0.9974650
10 1.457602e-03 0.9985424
11 1.180378e-03 0.9988196
12 2.614513e-02 0.9738549

$x
    
           LD1
  1  1.9897887
  2  2.4793572
  3  1.2519606
  4  2.7829001
  5  2.2210002
  6  1.6602417
  7  0.9698486
  8  1.1543287
  9  1.7957185
  10 1.9331885
  11 1.9855576
  12 1.2112739