+
aveweal
ra
naan orm wanna
panuohegenepraneanagme SADRReataser
soraEEN ELTENNnanTe
- With only one model‘of each type, which would ‘therefore
rigorously give only .one point on a behavior curve,
would be required
six shots
(including one repeat for an initiator test),
unless’ some combinations of the six items are possible.
Certain
combinations are possible, at sacrifice of definiteness, but it
| will be shown by the detailed considerations of the specific
plan that the minimum number of shots is three,
which has to be "good" in a sense to be defined.
each one of
When it is”
_ realized that atomic weapons are subject to fluctuations in be-
havior beyond human control, even a three-shot program appears
thin.
A test plan involving a small number of shots has to
ipnore such statistical considerations and relyon a degree of
compensation provided cy the fact that an unfavorable fluctua-tion, resulting in low yield,
tion,
can still yield behavior informa-
especially from other measured quantities such as alpha,
the rate of rise of the reaction rate,
In fact, this is general -
ly true regardless of the cause ofa low yleld and emphasizes that
—5--f
CUIER (OUP AN
SIACIRIE, It