made by Wroblewski (Comptes rendus, ci. 160), and by
Cailletet and Bouty (Journ. de phys. 1885, p. 297). The former’s
experiments were undertaken to test the suggestion made by
Clausius that the resistivity of pure metals is sensibly proportional
to the absolute temperature; he worked with copper having a
conductibility of 98%, and carried out measurements at various
temperatures, the lowest of which was that given by liquid
nitrogen boiling under reduced pressure. His general conclusion
was that the resistivity decreases much more quickly than the
absolute temperature, so as to approach zero at a point not far
below the temperature of nitrogen evaporating in vacuo. Cailletet
and Bouty, using ethylene as the refrigerant, and experimenting
at temperatures ranging from 0° C. to −100° C. and −123° C.,
constructed formulae intended to give the coefficients of variation
in electrical resistance for mercury, tin, silver, magnesium,
aluminium, copper, iron and platinum. Between 1892 and 1896
Dewar and Fleming carried out a large number of experiments
to ascertain the changes of conductivity that occur in metals
and alloys cooled in liquid air or oxygen to −200° C. The method
employed was to obtain the material under investigation in the
form of a fine regular wire and to wind it in a small coil; this
was then plunged in the liquid and its resistance determined.
The accompanying chart (fig. 14) gives the results in a compendious
form, the temperatures being expressed not in degrees
of the ordinary air-thermometer scale, but in platinum degrees
as given by one particular platinum resistance thermometer
which was used throughout the investigation. A table showing
the value of these degrees in degrees centigrade according to
Dickson will be found in the Phil. Mag. for June 1898, p. 527;
to give some idea of the relationship, it may be stated here that
−100° of the platinum thermometer = −94°.2 C., −150° plat.
= −140°.78 C., and −200° plat. = −185°.53 C. In general, the
resistance of perfectly pure metals was greatly decreased by cold—so
much so that, to judge by the course of the curves on the
chart, it appeared probable that at the zero of absolute temperature
resistance would vanish altogether and all pure metals
become perfect conductors of electricity. This conclusion,
however, has been rendered very doubtful by subsequent
observations by Dewar, who found that with the still lower
temperatures attainable with liquid hydrogen the increases of
conductivity became less for each decrease of temperature, until
a point was reached where the curves bent sharply round and
any further diminution of resistance became very small; that is,
the conductivity remained finite. The reduction in resistance
of some of the metals at the boiling point of hydrogen is very
remarkable. Thus copper has only 1105th, gold 130th, platinum 135th
to 117th, silver 124th the resistance at melting ice, but iron is only
reduced to 18th part of the same initial resistance. Table XIV.
shows the progressive decrease of resistance for certain metals
and one alloy as the temperature is lowered from that of boiling
water down to that of liquid hydrogen boiling under reduced
pressure; it also gives the “vanishing temperature,” at which
the conductivity would become perfect if the resistance continued
to decrease in the same ratio with still lower temperatures,
the values being derived from the extrapolation curves of the
relation between resistance and temperature, according to
Callendar and Dickson. It will be seen that many of the substances
have actually been cooled to a lower temperature than
that at which their resistance ought to vanish.
Metals and Alloys with Temperature. (Dewar and Fleming.)
In the case of alloys and impure metals, cold brings about a much smaller decrease in resistivity, and the continuations of the curves at no time show any sign of passing through the zero point. The influence of the presence of impurities in minute quantities is strikingly shown in the case of bismuth. Various specimens of the metal, prepared with great care by purely chemical methods, gave in the hands of Dewar and Fleming some very anomalous results, appearing to reach at −80° C. a maximum of conductivity, and thereafter to increase in resistivity with decrease of temperature. But when the determinations were carried out on a sample of really pure bismuth prepared electrolytically, a normal curve was obtained corresponding to that given by other pure metals. As to alloys, there is usually some definite mixture of two pure metals which has a maximum resistivity, often greater than that of either of the constituents. It appears too that high, if not the highest, resistivity corresponds to possible chemical compounds of the two metals employed, e.g. platinum 33 parts with silver 66 parts = PtAg4; iron 80 with nickel 20 = Fe4Ni; platinum 80 with iridium 20 = IrPt4; and copper 70 with manganese 30 = Cu2Mn. The product obtained by adding a small quantity of one metal to another has a higher specific resistance than the predominant constituent, but the curve is parallel to, and therefore the same in shape as, that of the latter (cf. the curves for various mixtures of Al and Cu on the chart). The behaviour of carbon and of insulators like gutta-percha, glass, ebonite, &c., is in complete contrast to the metals,
