72
PARTISAN REVIEW
mon sense, but as dependent on and continuous with it. Statements
of observation are, of course, not indubitable.
If
in a certain inquiry
there is no agreement as to whether a given assertion does or
doe~
not state an observation, it thereby becomes a
non-observation state–
ment
which must be confirmed on the basis of
observation statements
concerning whiCh no disagreement exists. All statements of observation
cannot be questioned at once, but no statement is above question.
Every assertion of an empirical science must, according to logical
empiricism, be either the statement of some observation, that is, a
basis sentence, or it must
be
testable by observation. Similarly, every
descriptive term of empirical science must be either an observation
term or must be
reducible
to observation terms. Non-observation
terms may be reduced to observation terms by stating the test con–
ditions and observations by which a simple sentence using the non–
observation term can be tested.
If
the statements describing the
test include other non-observation terms, these terms must in tum
be reduced, until a reduction is reached in which the test con–
ditions and observations are described entirely by means of observation
statements. A non-observation term for which no such reduction is
possible is not empirical.
The principle of reducibility should clarify some remarks made
above concerning the empirical requirements for literary or art critic–
ism. Logical empiricism does not demand that every term of criticism
designate
some observable trait of the work of art. What it does re–
quire is that
every
term be
reducible
to observation. A criticism that
satisfies this requirement will make only verifiable critical statements.
A statement that does not meet this requirement says nothing con–
cerning the subject of the criticism.
By this criterion many scientific concepts which sensationalist
empiricism viewed as metaphorical become literal empirical terms.
It also affords a method of determining the precise empirical content
of non-observation terms of empirical science.
It
is important to note
that this principle of reducibility does not mean that the non-observa–
tion terms could be eliminated from the language of science without
any loss of content. When the term "potential" is reduced to an ob–
servation basis of pointer-readings on experimental apparatus, one
cannot say that the potential is
nothing but
such pointer-readings.
Failure to realize this is ope of the fallacies underlying the "philo–
sophical" speculations of Eddington. The reductions of logical em–
piricism state certain logical connections between terms and not–
as in Eddington's metaphysics-the ontological constitution of things.
The foregoing analysis points to the conclusion that there are
