Platinum silver 6.62 micro-volts per degree C.

Nickelin 28.5 micro-volts per degree C.

German silver 10.43 micro-volts per degree C.

Manganin (St. Lindeck) 1.5 micro-volts per degree C.

Mechanically, the platinum silver is weak, and is greatly affected as to its resistance by mechanical strains--in fact, Klemencic considers it the worst substance he examined from this point of view--a conclusion rather borne out by Mr. Glazebrook's experience with the British a.s.sociation standards already referred to (B. A. Reports, 1891 and 1892).

Taking everything into account, it will probably be well to construct standards either with oil insulation only, or to bake the coils in sh.e.l.lac before testing, instead of soaking in paraffin. Fig. 89 ill.u.s.trates a form of an oil immersed standard which is in use in my laboratory, and through which a considerable current may be pa.s.sed.

The oil is stirred by means of a screw propeller.

Fig. 89.

Fig. 89 represents a standard resistance for making Clerk cell comparisons by the silver voltameter method. The framework on which the coils are wound consists of a base and top of slate. The pillars are of flint gla.s.s tube surrounding bra.s.s bolts, and cemented to the latter by raw sh.e.l.lac. Grooves are cut in the gla.s.s sleeves to hold the wires well apart. These grooves were cut by means of a file working with kerosene lubrication. A screw stirrer is provided, and the whole apparatus is immersed in kerosene in the gla.s.s box of a storage cell. The apparatus is aged to begin with by heating to a temperature a good deal higher than any temperature it is expected to reach in actual work. After this the rigidity of the frame is intended to prevent any further straining of the wire. The apparatus as figured is not intended to be cooled to 0 C, so that it is put in as large a box as possible to gain the advantage of having a large volume of liquid. The top and bottom slates measure seven inches by seven inches, and the distance between them is seven inches. The inner coil is wound on first, and the loop which const.i.tutes the end of the winding is brought up to a suitable position for adjustment.

The insulation of the heavy copper connectors is by means of ebonite.

-- 120. Platinum Iridium.

Platinum 90 per cent, iridium 10 per cent. This material was prepared in some quant.i.ty at the cost of the French Government, and distributed for test about 1886. Klemencic got some of it as representing Austria, and found it behaved very like the platinum silver alloy just discussed. The temperature coefficient is, however, higher than for platinum silver (0.00126 as against 0.00027). The mechanical properties of the alloy are, however, much better than those of the silver alloy; and in view of the experience with B. A. standards above quoted, it remains an open question whether, on the whole, it would not be the better material for standards, in spite of its higher price. Improvements in absolute measurements of resistance, however, may render primary standards superfluous.

-- 121. Manganin.

Discovered by Weston--at all events as to its application to resistance coils. A glance at the diagram will exhibit its unique properties, on account of which it has been adopted by the Physikalisch Technischen Reichsanstalt for resistance standards. The composition of the alloy is copper 84 per cent, manganese 12 per cent, nickel 4 per cent, and it is described as of a steel-gray colour.

Unfortunately it is apt to oxidise in the air, or rather the manganese it contains does so, so that it wants a very perfect protection against the atmosphere.

Like German silver, manganin changes in resistance on winding, and coils made of it require to be artificially aged by heating to 150 for five hours before final adjustment. The annealing cannot be carried out in air, owing to the tendency to oxidation. The method adopted by St. Lindeck (at all events up to 1892) is to treat the coil with thick alcoholic sh.e.l.lac varnish till the insulation is thoroughly saturated, and then to bake the coil as described. The baking not only anneals the wire, but reduces the sh.e.l.lac to a hard and highly insulating ma.s.s.

Whether stresses of sufficient magnitude to produce serious mechanical effects can be set up by unequal expansion of wire and sh.e.l.lac during heating and cooling is not yet known, but so far as tested (and it must be presumed that the Reichsanstalt tests are thorough) no difficulty seems to have been met with. In course of time, however, probably the best sh.e.l.lac coating will crack, and then adieu to the permanency of the coil! This might, of course, be obviated by keeping the coil in kerosene, which has no action on sh.e.l.lac, but which decomposes somewhat itself.

The method of treatment above described suffices to render coils of manganin constant for at least a year (in 1892 the tests had only been made for this time) within a few thousands per cent. Manganin can be obtained in sheets, and from this material standards of 10-2, 10-3, and 10-4 ohms are made by soldering strips between stout copper bars, and these are adjusted by gradually increasing their resistance by boring small holes through them. The solder employed is said to be "silver."

Mr. Griffiths (Phil. Trans. vol. clx.x.xiv. [1893], A, p. 390) has had some experience with manganin carrying comparatively heavy currents, under which circ.u.mstances its resistance when immersed in water was found to rise in spite of the varnish which coated it.

Other experiments in which the manganin wire was immersed in paraffin oil did not exhibit this effect, though stronger currents were pa.s.sed.

On the whole, manganin appears to be the best material for coil boxes and "secondary" resistance standards. Whether it is fit to rank with the platinum alloys as regards permanency must be treated as an open question.

-- 122. Other Alloys.

The following tables, taken from the work of Feussner and St.

Lindeck, Zeitschrift fuer Instrumenten Kunde, 1889, vol. ix. p.

233, together with the following notes, will suffice.

-- 123. Nickelin.

This is only German silver with a little less zinc, a little more nickel, and traces of cobalt and manganese. It behaves like German silver, but is an improvement on the latter in that all the faults of German silver appear upon a reduced scale in nickelin.

-- 124. Patent Nickel.

Practically a copper nickel alloy, used to some extent by Siemens and Halske. It stands pretty well in the same relation to nickelin as the latter does to German silver. After annealing as for manganin it can be made into serviceable standards which do not change more than a few thousandths per cent. I have not come across a statement of its thermo-voltage against copper.

-- 125. Constantin.

Another nickel copper alloy containing 50 per cent of each const.i.tuent. It appears to be a serviceable substance, having a temperature coefficient of 0.003 per cent per degree only, but an exceedingly high thermo-voltage, viz. 40 micro-volts per degree against copper.

1 2 3 4 5 6 7 8 German Nickelin made Rheo- Patent Nickel Manga- Nickel Silver by Obermaier tane nese Manga- Dia- Dia- Dia- Dia- Copper nese meter meter meter meter Copper 1.0mm 0.1mm 0.6mm 1.0mm

Copper 60.16 61.63 54.57 53.28 74.41 74.71 70 73

Zinc 25.37 19.67 20.44 16.89 0.23 0.52 ... ...

Tin ... ... ... ... trace ... ...

Nickel 14.03 18.46 24.48 25.31 25.10 24.14 ... 3

Iron 0.30 0.24 0.64 4.46 0.42 0.70 ... ...

Cobalt trace 0.19 ... ... trace trace ... ...

Mang- trace 0.18 0.27 0.37 0.13 0.17 30 24 anese.

99.86 100.37 100.40 100.31 100.24 100.24 ... ...

Specific resistance 30.0 33.2 44.8 52.5 34.2 32.8 100.6 47.7

Temperature coefficient 0.00036 0.00030 0.00033 0.00041 0.00019 0.00021 0.00004 0.00003

The specific resistance is in microhms, i.e. 10-6 ohms per cubic centimetre, and the temperature coefficient in degrees centigrade.

126. Nickel Manganese Copper.

I can find no other reference with regard to this alloy mentioned by Lindeck. Nicholls, however (Silliman's Journal [3], 39, 171, 1890), gives some particulars of alloys of copper and ferromanganese. The following table is taken from Wiedemann's Beiblatter (abstract of Nicholl's paper, 1890, p. 811). All these alloys appear to require annealing at a red heat before their resistances are anything like constant.

Let x be percentage of copper, then 100--x is percentage of "ferromanganese."

Values of x. 100 99.26 91 .88 86.98 80.4 70.65

Specific resistance with respect to copper (? pure) 1 1.19 11.28 20.4 27.5 45.1

Temperature coefficient per degree x 10^6(hard) 3202 2167 138 16 22 -24

Ditto (soft) ... ... 184 80 66 21

If nickel is added, alloys of much the same character are obtained, some with negative temperature coefficients--for instance, one containing 52.51 per cent copper, 31.27 per cent ferromanganese, and 16.22 nickel.

A detailed account of several alloys will be found in a paper by Griffiths (Phil. Trans. 1894, p. 390), but as the constants were determined to a higher order of accuracy than the composition of the material--or, at all events, to a higher degree of accuracy than that to which the materials can be reproduced--there is no advantage in quoting them here.