106 VALUATION, DEPRECIATION AND THE RATE-BASE
service plant, and to the difficulty of determining the expect-
ancy of those articles which have already been in use for
some time, it is nevertheless important that the whole ques-
tion be fully considered in order that a framework may be
constructed into which the best data furnished by experience
can be fitted.
The non-agreement of actual life of individual items with
their probable life and the extent to which this lack of agree-
ment should be taken into account in estimating present worth
and in estimating replacement requirements, in line with this
thought, has been studied on various assumptions with inter-
esting results. These will be briefly referred to, and the re-
sulting tables are presented for use until, in the light of larger
experience, they can be replaced with better ones.
Assumptions Relating to Departure of Actual from Probable
Life. — Unfortunately there are no records available from which
absolutely dependable tables of expectancy could be prepared
for each class of perishable articles in use in connection with
public service properties, such as have been prepared by actu-
aries for human beings. Any assumption in this regard is more
or less conjecture. Nevertheless, it is interesting to note what
the expectancy would be at various ages, if certain definite
reasonable assumptions are made.
When any large number of articles which have the same
probable life, as, for example, ten years, is under consideration,
there will be as many service years in the aggregate, represented
by the failures to reach the probable life term of ten years, as
there will be service years represented by those articles which
outlast the ten-year term. It may also be accepted as a cer-
tainty that there will be a greater number of articles per year
to go out of use in the years just preceding and just following
the term limit than at any other time. This suggests conform-
ity with the law of probabilities.
All articles in a group having a ten-year probable life might
fail (a) exactly at the end of ten years, but this is highly improb-
able. The individual articles might fail (0) at a uniform rate
