W
_
W(A,A2)
Nx~
N2
N
=
W(A2A3)
N
i
AT
2
or
w W\a2A3)
Thus, if the
measuring
wire has
a
constant
cross
section and is
homogeneous,
we
also have
W
_
*
i
"
F7*
Thus,
three [different]
readings eliminate
the
source
of
error.
During
the measurement the current must be of short
duration,
otherwise the
heating
due to the current flow would
considerably
increase the
resistance.
The
relation
between specific resistance
and
temperature.
One must here
distinguish
between
a
dependence
on
the
physical
and
on
the
chemical
structure of the
metal.
The former
play[s]
a
secondary
role in
comparison
with the latter.
Pure metals
obey
a
simple law. If
one
makes
a
series of
resistance measurements at various temperatures for
a
wire of
a
completely
pure
substance,
one
obtains
a
strictly
linear relation
between
W
and
t
W
t =
W0(1
+
at)
-W0
a
'
Thus, it
suffices to conduct the measurement at temperatures
O
and t.
Different
metals
yield
a
between
0.0038 and 0.0042.
This value
roughly
agrees
with the coefficient found
in the
investigation
of the
relationship
betw[een] temperature
and
pressure
2
of
an
ideal
gas.
(273)
P
=
Po(1
+
at). Thus, it
turns out that the
conductivity of
pure
metals would become
«
large
at the
same
temperature at
which the
gas pressure
becomes 0,,
provided that
the
linear
relationship still
holds
for low
temperatures. The
assumption
that this is indeed
so was
confirmed in
an
investigation
of the
resistance of metals at the temperature of
liquid atmospheric
air
(-192°). Thus, it is almost certain that the
posited 0-point
of
temperature corresponds to reality.
Alloys
behave
completely differently.
Even
though
the absolute
value of
their
specific
resistance is
considerably higher
than that of
pure
metals,
it varies much less with the temperature, often
only
1%.
111