In building a sound theory there are four steps: 
(a) Clearly define the problem. This may not be what 
it first appears: be sure you know what the difficulty is. 
( b ) Cast about for possible solutions—not only the 
first one which occurs to you, but as many as you can 
figure out; then compare their merits and select the most 
probable one. 
(c) Reason out the developments of this idea to its 
logical conclusion. 
(d) Constantly test your theory by searching for further 
evidence or by experiment. Keep your eyes open for 
evidence indicating some other theory as more probable 
and give honest weight to it. 
(2) The complete case. —To be complete, the case must 
answer the following questions: (a) What was the offense? 
(6) Where was it committed? (c) When was it committed? 
(d) How was it accomplished? (e) Who did it? (/) Why 
did he do it? 
Memorize these six words: what, where, when, how, 
ivho, why, and frequently test by them the completeness 
of both your theory and the facts so far actually estab¬ 
lished. This will be one of the greatest helps in planning 
what remains to be done. 
(3) The case which will stand in court. —Proof which will 
convince a jury “beyond a reasonable doubt,” and which 
is necessary for a criminal trial, is much more difficult 
to establish than a case which will satisfy the investi¬ 
gator. Individual judgments take much knowledge 
for granted, but a court must, generally speaking, 
have actual proof of every material point. If you hear 
a shot, for example, and on going in the direction of the 
sound find a man standing over a dead doe, you will 
not be long in reaching a conclusion. But if you arrest 
him while he is only looking at the deer, you are liable 
to lose your case in court. You might possibly be able 
to find some one who saw him shoot and the doe fall. 
If he takes possession, however, you have him on that 
count, whether you can prove that he killed it or not 
Whenever a fact is found which points to a material con¬ 
clusion, ask yourself: (a) Does this sufficiently prove the 
