Section 28. (f) Probability.
Inasmuch as the work of the criminal judge depends upon the
proof of evidence, it is conceivable that the thing for him most
important is that which has evidential character or
force.[1] A sufficient
definition of evidence or proof does not exist because no bounds
have been set to the meaning of "Proved." All disciplines furnish
examples of the fact that things for a long time had probable validity,
later indubitable validity; that again some things were considered
proved and were later shown to be incorrect, and that many things at
one time wobbly are in various places, and even among particular
persons, supposed to be at the limits of probability and proof.
Especially
remarkable is the fact that the concept of
the
proved is very
various in various sciences, and it would be absorbing to establish
the difference between what is called proved and what only probable
in a number of given examples by the mathematician, the physicist,
the chemist, the physician, the naturalist, the philologist, the historian,
the philosopher, the lawyer, the theologian, etc. But this is no
task for us and nobody is called upon to determine who knows what
"Proved" means. It is enough to observe that the differences are
great and to understand why we criminalists have such various
answers to the question: Is this proved or only probable? The
varieties may be easily divided into groups according to the mathematical,
philosophic, historical or naturalistic inclinations of the
answerer. Indeed, if the individual is known, what he means by
"proved" can be determined beforehand. Only those minds that
have no especial information remain confused in this regard, both
to others and to themselves.
Sharply to define the notion of "proved" would require at least
to establish its relation to usage and to say: What we desire leads
us to an assumption, what is possible gives us
probability, what
appears certain, we call proved. In this regard
the second is always,
in some degree, the standard for the first (desires, e. g., cause us to
act; one becomes predominant and is fixed as an assumption which
later on becomes clothed with a certain amount of reliability by
means of this fixation).
The first two fixations, the assumption and the probability, have
in contrast to their position among other sciences only a heuristic
interest to us criminalists. Even assumptions, when they become
hypotheses, have in various disciplines a various value, and the
greatest lucidity and the best work occur mainly in the quarrel about
an acutely constructed hypothesis.
Probability has a similar position in the
sciences. The scholar
who has discovered a new thought, a new order, explanation or
solution, etc., will find it indifferent whether he has made it only
highly probable or certain. He is concerned only with the idea, and
a scholar who is dealing with the idea for its own sake will perhaps
prefer to bring it to a great probability rather than to indubitable
certainty, for where conclusive proof is presented there is no longer
much interest in further research, while probability permits and
requires further study. But our aim is certainty and proof only,
and even a high degree of probability is no better than untruth and
can not count. In passing judgment and for the purpose of judgment
a high degree of probability can have only corroborative weight,
and then it is probability only when taken in itself, and proof
when taken with regard to the thing it corroborates. If, for example,
it is most probable that X was recognized at the place of a crime,
and if at the same time his evidence of alibi has failed, his footmarks
are corroborative; so are the stolen goods which have been seen
in his possession, and something he had lost at the place of the crime
which is recognized as his property, etc. ln short, when all these
indices are in themselves established only as highly probable, they
give under certain circumstances, when taken together, complete
certainty, because the coincidence of so many high probabilities must
be declared impossible if X were not the criminal.
In all other cases, as we have already pointed
out, assumption
and probability have only a heuristic value for us lawyers. With
the assumption, we must of course count; many cases can not be
begun without the assistance of assumption. Every only
half-confused case, the process of which is unknown, requires first of
all and as early as possible the application of some assumption to
its material. As soon as the account is inconsistent the assumption
must be abandoned and a fresh one and yet again a fresh one assumed,
until finally one holds its own and may be established as probable.
It then remains the center of operation, until it becomes of itself a
proof or, as we have explained, until so many high probabilities in
various directions have been gathered, that, taken in their order, they
serve evidentially. A very high degree of probability is sufficient
in making complaints; but sentencing requires "certainty," and in
most cases the struggle between the prosecution and the defense,
and the doubt of the judge, turns upon the question of probability
as against proof.[2]
That probability is in this way and in a number of relations, of
great value to the criminalist can not appear doubtful. Mittermaier
defines its significance briefly: "Probability naturally can
never lead to sentence. It is, however, important as a guide for the
conduct of the examiner, as authorizing him to take certain measures;
it shows how to attach certain legal processes in various directions."
Suppose that we review the history of the development of the
theory of probability. The first to have attempted a sharp distinction
between demonstrable and probable knowledge was Locke.
Leibnitz was the first to recognize the importance of the theory
of probability for inductive logic. He was succeeded by the mathematician
Bernoulli and the revolutionist Condorcet. The theory in
its modern form was studied by Laplace, Quetelet, Herschel, von
Kirchmann, J. von Kries, Venn, Cournot, Fick, von Bortkiewicz, etc.
The concept that is called probability varies with different authorities.
Locke
[3] divides all
fundamentals into demonstrative and probable.
According to this classification it is probable that "all men are mortal,"
and that "the sun will rise to-morrow." But to be consistent
with ordinary speech the fundamentals must be classified as evidence,
certainties, and probabilities. By certainties I understand such
fundamentals as are supported by experience and leave no room
for doubt or consideration—everything else, especially as it permits
of further proof, is more or less probable.
Laplace[4] spoke more
definitely—"Probability depends in part
on our ignorance, in part on our knowledge . . .
"The theory of probability consists in the reduction of doubts
of the same class of a definite number of equally possible cases in
such a way that we are equally undetermined with regard to their
existence, and it further consists in the determination of the number
of those cases which are favorable to the result the probability of
which is sought. The relation of this number to the number of all
possible cases is the measure of the probability. It is therefore a
fraction the numerator of which is derived from the number of
cases favorable to the result and the denominator from the number
of all possible cases." Laplace, therefore, with J. S. Mill, takes
probability to be a low degree of certainty, while
Venn[5] gives it an
objective support like truth. The last view has a great deal of
plausibility inasmuch as there is considerable doubt whether an
appearance is to be taken as certain or as only probable. If this
question is explained, the assertor of certainty has assumed some
objective foundation which is indubitable at least subjectively.
Fick represents the establishment of probability as a fraction as
follows: "The probability of an incompletely expressed hypothetical
judgment is a real fraction proved as a part of the whole universe
of conditions upon which the realization of the required result
necessarily depends.
"According to this it is hardly proper to speak of the probability
of any result. Every individual event is either absolutely necessary
or impossible. The probability is a quality which can pertain only
to a hypothetical judgment."
[6]
That it is improper to speak of the probability of a result admits
of no doubt, nor will anybody assert that the circumstance of
to-morrow's rain is in itself probable or improbable—the form of
expression is only a matter of usage. It is, however, necessary to
distinguish between conditioned and unconditioned probability.
If I to-day consider the conditions which are attached to the ensuing
change of weather, if I study the temperature, the barometer, the
cloud formation, the amount of sunlight, etc., as conditions which
are related to to-morrow's weather as its forerunners, then I must
say that to-morrow's rain is probable to such or such a degree. And
the correctness of my statement depends upon whether I know
the conditions under which rain must appear, more
or less accurately
and completely, and whether I relate those conditions properly.
With regard to unconditioned probabilities which have nothing to
do with the conditions of to-day's weather as affecting to-morrow's,
but are simply observations statistically made concerning the
number of rainy days, the case is quite different. The distinction
between these two cases is of importance to the criminalist because
the substitution of one for the other, or the confusion of one with the
other, will cause him to confuse and falsely to interpret the probability
before him. Suppose, e. g., that a murder has happened in
Vienna, and suppose that I declare immediately after the crime and
in full knowledge of the facts, that according to the facts, i. e., according
to the conditions which lead to the discovery of the criminal,
there is such and such a degree of probability for this discovery.
Such a declaration means that I have calculated a conditioned probability.
Suppose that on the other hand, I declare that of the
murders occurring in Vienna in the course of ten years, so and
so many are unexplained with regard to the personality of the
criminal, so and so many were explained within such and such a
time,—and consequently the probability of a discovery in the case
before us is so and so great. In the latter case I have spoken of
unconditioned probability. Unconditioned probability may be
studied by itself and the event compared with it, but it must never
be counted on, for the positive cases have already been reckoned
with in the unconditioned percentage, and therefore should not be
counted another time. Naturally, in practice, neither form of
probability is frequently calculated in figures; only an approximate
interpretation of both is made. Suppose that I hear of a certain
crime and the fact that a footprint has been found. If without
knowing further details, I cry out: "Oh! Footprints bring little
to light!" I have thereby asserted that the statistical verdict in
such cases shows an unfavorable percentage of unconditional probability
with regard to positive results. But suppose that I have
examined the footprint and have tested it with regard to the other
circumstances, and then declared: "Under the conditions before
us it is to be expected that the footprint will lead to results"—
then I have declared, "According to the conditions the conditioned
probability of a positive result is great." Both assertions may be
correct, but it would be false to unite them and to say, "The conditions
for results are very favorable in the case before us, but
generally hardly anything is gained by means of footprints, and
hence the probability in this case is small." This would be false
because the few favorable results as against the many unfavorable
ones have already been considered, and have already determined
the percentage, so that they should not again be used.
Such mistakes are made particularly when determining the complicity
of the accused. Suppose we say that the manner of the
crime makes it highly probable that the criminal should be a
skilful, frequently-punished thief, i. e., our probability is conditioned.
Now we proceed to unconditioned probability by saying: "It is
a well-known fact that frequently-punished thieves often steal
again, and we have therefore two reasons for the assumption that
X, of whom both circumstances are true, was the criminal." But
as a matter of fact we are dealing with only one identical probability
which has merely been counted in two ways. Such inferences are
not altogether dangerous because their incorrectness is open to
view; but where they are more concealed great harm may be done
in this way.
A further subdivision of probability is made by
Kirchmann.[7]
He distinguished:
(1) General probability, which depends upon the causes or consequences
of some single uncertain result, and derives its character
from them. An example of the dependence on causes is the collective
weather prophecy, and of dependence on consequences is Aristotle's
dictum, that because we see the stars turn the earth must
stand still. Two sciences especially depend upon such probabilities:
history and law, more properly the practice and use of criminal
law. Information imparted by men is used in both sciences, this
information is made up of effects and hence the occurrence is inferred
from as cause.
(2) Inductive probability. Single events which must be true,
form the foundation, and the result passes to a valid universal.
(Especially made use of in the natural sciences, e. g., in diseases
caused by bacilli; in case X we find the appearance A and in diseases
of like cause Y and Z, we also find the appearance A. It is therefore
probable that all diseases caused by bacilli will manifest the symptom
A.)
(3) Mathematical Probability. This infers that A is connected
either with B or C or D, and asks the degree of probability. I. e.:
A woman is brought to bed either with a boy or a girl: therefore
the probability that a boy will be born is one-half.
Of these forms of probability the first two are of equal importance
to us, the third rarely of value, because we lack arithmetical
cases and because probability of that kind is only of transitory worth
and has always to be so studied as to lead to an actual counting of
cases. It is of this form of probability that Mill advises to know,
before applying a calculation of probability, the necessary facts,
i. e., the relative frequency with which the various events occur, and
to understand clearly the causes of these events. If statistical
tables show that five of every hundred men reach, on an average,
seventy years, the inference is valid because it expresses the existent
relation between the causes which prolong or shorten life.
A further comparatively self-evident division is made by Cournot,
who separates subjective probability from the possible probability
pertaining to the events as such. The latter is objectively
defined by Kries[8] in the following example:
"The throw of a regular die will reveal, in the great majority of
cases, the same relation, and that will lead the mind to suppose it
objectively valid. It hence follows, that the relation is changed
if the shape of the die is changed." But how "this objectively valid
relation," i. e., substantiation of probability, is to be thought of,
remains as unclear as the regular results of statistics do anyway.
It is hence a question whether anything is gained when the form of
calculation is known.
Kries says, "Mathematicians, in determining the laws of probability,
have subordinated every series of similar cases which take
one course or another as if the constancy of general conditions, the
independence and chance equivalence of single events, were identical
throughout. Hence, we find there are certain simple rules according
to which the probability of a case may be calculated from the number
of successes in cases observed until this one and from which,
therefore, the probability for the appearance of all similar cases
may be derived. These rules are established without any exception
whatever." This statement is not inaccurate because the general
applicability of the rules is brought forward and its use defended
in cases where the presuppositions do not agree. Hence, there are
delusory results, e. g., in the calculation of mortality, of the statements
of witnesses and judicial deliverances. These do not proceed
according to the schema of the ordinary play of accident. The
application, therefore, can be valid only if the constancy of general
conditions may be reliably assumed.
But this evidently is valid only with regard to unconditioned
probability which only at great intervals and transiently may
influence our practical work. For, however well I may know that
according to statistics every xth witness is punished for perjury, I
will not be frightened at the approach of my xth witness though
he is likely, according to statistics, to lie. In such cases we are not
fooled, but where events are confused we still are likely to forget
that probabilities may be counted only from great series of figures
in which the experiences of individuals are quite lost.
Nevertheless figures and the conditions of figures with regard
to probability exercise great influence upon everybody; so great indeed,
that we really must beware of going too far in the use of figures.
Mill cites a case of a wounded Frenchman. Suppose a regiment
made up of 999 Englishmen and one Frenchman is attacked and
one man is wounded. No one would believe the account that this
one Frenchman was the one wounded. Kant says significantly:
"If anybody sends his doctor 9 ducats by his servant, the doctor
certainly supposes that the servant has either lost or otherwise
disposed of one ducat." These are merely probabilities which
depend upon habits. So, it may be supposed that a handkerchief
has been lost if only eleven are found, or people may wonder at the
doctor's ordering a tablespoonful every five quarters of an hour,
or if a job is announced with $2437 a year as salary.
But just as we presuppose that wherever the human will played
any part, regular forms will come to light, so we begin to doubt
that such forms will occur where we find that accident, natural
law, or the unplanned coöperation of men were determining factors,
If I permit anybody to count up accidentally concurrent things
and he announces that their number is one hundred, I shall probably
have him count over again. I shall be surprised to hear that somebody's
collection contains exactly 1000 pieces, and when any one
cites a distance of 300 steps I will suppose that he had made an
approximate estimation but had not counted the steps. This fact
is well known to people who do not care about accuracy, or who
want to give their statements the greatest possible appearance of
correctness; hence, in citing figures, they make use of especially
irregular numbers, e. g. 1739, 7/8, 3.25%, etc. I know a case of a
vote of jurymen in which even the proportion of votes had to be
rendered probable. The same jury had to pass that day on three
small cases. In the first case the proportion was 8 for, 4 against,
the second case showed the same proportion and the third case the
same. But when the foreman observed the proportion he announced
that one juryman must change his vote because the same proportion
three times running would appear too improbable! If we want
to know the reason for our superior trust in irregularity in such
cases, it is to be found in the fact that experience shows nature, in
spite of all her marvelous orderliness in the large, to be completely
free, and hence irregular in little things. Hence, as Mill shows in
more detail, we expect no identity of form in nature. We do not
expect next year to have the same order of days as this year, and
we never wonder when some suggestive regularity is broken by a
new event. Once it was supposed that all men were either black
or white, and then red men were discovered in America. Now
just exactly such suppositions cause the greatest difficulties, because
we do not know the limits of natural law. For example, we do not
doubt that all bodies on earth have weight. And we expect to find
no exception to this rule on reaching some undiscovered island on
our planet; all bodies will have weight there as well as everywhere
else. But the possibility of the existence of red men had to be granted
even before the discovery of America. Now where is the difference
between the propositions: All bodies have weight, and, All men are
either white or black? It may be said circularly the first is a natural
law and the second is not. But why not? Might not the human
body be so organized that according to the natural law it would be
impossible for red men to exist? And what accurate knowledge
have we of pigmentation? Has anybody ever seen a green horse?
And is the accident that nobody has ever seen one to prevent the
discovery of green horses in the heart of Africa? May, perhaps,
somebody not breed green horses by crossings or other experiments?
Or is the existence of green horses contrary to some unknown but
invincible natural law? Perhaps somebody may have a green horse
to-morrow; perhaps it is as impossible as water running up hill.
To know whether anything is natural law or not always depends
upon the grade and standing of our immediate experience—and
hence we shall never be able honestly to make any universal proposition.
The only thing possible is the greatest possible accurate
observation of probability in all known possible cases, and of the
probability of the discovery of exceptions. Bacon called the establishment
of reliable assumptions, counting up without meeting any
contradictory case. But what gives us the law is the manner of
counting. The untrained mind accepts facts as they occur without
taking the trouble to seek others; the trained mind seeks the facts
he needs for the premises of his inference. As Mill says, whatever
has shown itself to be true without exception may be held universal
so long as no doubtful exception is presented, and when the case
is of such a nature that a real exception could not escape our observation.
This indicates how we are to interpret information given by
others. We hear, "Inasmuch as this is always so it may be assumed
to be so in the present case." Immediate acceptance of this proposition
would be as foolhardy as doubt in the face of all the facts.
The proper procedure is to examine and establish the determining
conditions, i. e., who has counted up this "always," and what caution
was used to avoid the overlooking of any exception. The real
work of interpretation lies in such testing. We do not want to reach
the truth with one blow, we aim only to approach it. But the step
must be taken and we must know how large it is to be, and know
how much closer it has brought us to the truth. And this is learned
only through knowing who made the step and how it was made.
Goethe's immortal statement, "Man was not born to solve the
riddle of the universe, but to seek out what the problem leads to
in order to keep himself within the limits of the conceivable," is
valid for us too.
Our great mistake in examining and judging often lies in our
setting too much value upon individual circumstances, and trying
to solve the problem with those alone, or in not daring to use any
given circumstance sufficiently. The latter represents that stupidity
which is of use to scientific spirits when they lack complete proof
of their points, but is dangerous in practical affairs. As a rule, it is
also the consequence of the failure to evaluate what is given, simply
because one forgets or is too lazy to do so. Proper action in this
regard is especially necessary where certain legal proceedings have
to occur which are entitled to a definite degree of probability without
requiring certainty, i. e., preliminary examinations, arrests,
investigations of the premises, etc. No law says how much probability
is in such cases required. To say how much is impossible, but it
is not unwise to stick to the notion that the event must appear
true, if not be proved true, i. e., nothing must be present to destroy
the appearance of truth. As Hume says, "Whenever we have reason
to trust earlier experiences and to take them as standards of
judgment of future experiences, these reasons may have probability."
The place of probability in the positive determination of the
order of modern criminal procedure is not insignificant. When the
law determines upon a definite number of jurymen or judges, it
is probable that this number is sufficient for the discovery of the
truth. The system of prosecution establishes as a probability that
the accused is the criminal. The idea of time-lapse assumes the
probability that after the passage of a certain time punishment
becomes illusory, and prosecution uncertain and difficult. The
institution of experts depends on the probability that the latter
make no mistakes. The warrant for arrest depends on the probability
that the accused behaved suspiciously or spoke of his crime,
etc. The oath of the witness depends on the probability that the
witness will be more likely to tell the truth under oath, etc.
Modern criminal procedure involves not only probabilities but
also various types of possibility. Every appeal has for its foundation
the possibility of an incorrect judgment; the exclusion of certain
court officials is based on the possibility of prejudice, or at least
on the suspicion of prejudice; the publicity of the trial is meant to
prevent the possibility of incorrectness; the revision of a trial
depends on the possibility that even legal sentences may be false
and the institution of the defendant lawyer depends upon the possibility
that a person without defense may receive injustice. All
the formalities of the action of the court assume the possibility
that without them improprieties may occur, and the institution of
seizing letters and messages for evidence, asserts only the possibility
that the latter contain things of importance, etc.
When the positive dicta of the law deal with possibility and
probability
in questions of great importance the latter become especially
significant.
We have yet to ask what is meant by "rule" and what its relation
is to probability. Scientifically "rule" means law subjectively
taken and is of equal significance with the guiding line for
one's own conduct, whence it follows that there are only rules of
art and morality, but no rules of nature. Usage does not imply
this interpretation. We say that as a rule it hails only in the daytime;
by way of exception, in the night also; the rule for the appearance
of whales indicates that they live in the Arctic Ocean;
a general rule indicates that bodies that are especially soluble in
water should dissolve more easily in warm than in cold water, but
salt dissolves equally well in both. Again we say: As a rule the
murderer is an unpunished criminal; it is a rule that the brawler is
no thief and vice versa; the gambler is as a rule a man of parts,
etc. We may say therefore, that regularity is equivalent to customary
recurrence and that whatever serves as rule may be expected
as probable. If, i. e., it be said, that this or that happens as a rule,
we may suppose that it will repeat itself this time. It is not permissible
to expect more, but it frequently happens that we mistake
rules permitting exceptions for natural laws permitting none. This
occurs frequently when we have lost ourselves in the regular occurrences
for which we are ourselves responsible and suppose that
because things have been seen a dozen times they must always
appear in the same way. It happens especially often when we have
heard some phenomenon described in other sciences as frequent and
regular and then consider it to be a law of nature. In the latter case
we have probably not heard the whole story, nor heard general
validity assigned to it. Or again, the whole matter has long since
altered. Lotze wrote almost half a century ago, that he had some
time before made the statistical observation that the great positive
discoveries of exact physiology have an average life of about four
years. This noteworthy statement indicates that great positive
discoveries are set up as natural laws only to show themselves as
at most regular phenomena which have no right to general validity.
And what is true of physiology is true of many other sciences, even
of the great discoveries of medicine, even legal medicine. This,
therefore, should warn against too much confidence in things that
are called "rules." False usage and comfortable dependence upon
a rule have very frequently led us too far. Its unreliability is shown
by such maxims as "Three misses make a rule" or "Many stupidities
taken together give a golden rule of life," or "To-day's exception
is to-morrow's rule," or the classical perversion: "The rule that
there are no rules without exception is a rule without exception,
hence, there is one rule without exception."
The unreliability of rules is further explained by their rise from
generalization. We must not generalize, as Schiel says, until we
have shown that if there are cases which contradict our generalizations
we know those contradictions. In practice approximate generalizations
are often our only guides. Natural law is too much
conditioned, cases of it too much involved, distinctions between
them too hard to make, to allow us to determine the existence of a
natural phenomenon in terms of its natural characteristics as a
part of the business of our daily life. Our own age generalizes
altogether too much, observes too little, and abstracts too rapidly.
Events come quickly, examples appear in masses, and if they are
similar they tend to be generalized, to develop into a rule, while the
exceptions which are infinitely more important are unobserved, and
the rule, once made, leads to innumerable mistakes.
[[ id="n28.1"]]
B. Petronievics: Der Satz vom Grunde. Leipzig 1898.
[[ id="n28.2"]]
Of course we mean by "proof" as by "certainty" only the highest possible
degree of probability.
[[ id="n28.3"]]
Locke: Essay on the Human Understanding.
[[ id="n28.4"]]
Laplace: Essay Philosophique sur les Probabilités. Paris 1840.
[[ id="n28.5"]]
Venn: The Logic of Chance.
[[ id="n28.6"]]
Philos. Versuch über die Wahrscheinlichkeiten. Würzburg 1883.
[[ id="n28.7"]]
Über die Wahrscheinlichkeit, Leipzig 1875.
[[ id="n28.8"]]
J. v. Kries: Über die Wahrseheinlichkeit Il. Möglichkeit u. ihre
Bedeutung in
Strafrecht. Zeitschrift f. d. ges. St. R. W. Vol. IX, 1889.