Resistance pyrometers made by other firms differ in detail from the foregoing. In the Siemens pyrometer the coil is wound on special fireclay, and protected by an iron sheath, the space between the coil and the sheath being filled with magnesia, which effectively prevents the corrosion of the platinum; and compensation is effected by means of a single wire passing down the centre and connected to one end of the coil, a special form of Wheatstone bridge being used to take the measurement. In the instruments made by R. W. Paul the coil is made of flat strip rolled out from wire, wound on mica, and protected by a silica tube and outer iron sheath. The Leeds-Northrup Company of Philadelphia employ a rod of obsidian on which to wind the coil, and also make a form in which the coil is wound so as to be self-sustaining, thus dispensing with the support. In all cases the coil is wound non-inductively, _i.e._ the wire is doubled before making into a spiral. The zero resistance of a given instrument depends upon the accuracy of the measuring appliances used, and upon the degree of precision it is desired to attain. If, for example, it is intended to read to 1°C., with appliances capable of measuring to 1/100 of an ohm, a convenient zero resistance is 2·6 ohms; it being found that with pure platinum the resistance rises from 2·6 ohms at 0° to 3·6 ohms at 100° C., an increase of 1/100 of an ohm for 1° C. With coarser measuring arrangements, for the same degree of precision, a correspondingly higher zero resistance will be required; thus if 1/25 ohm be the least amount detectable by the measuring device, a zero resistance of 10·4 ohms would enable 1° C. to be observed. It is evident that a suitable zero resistance may be calculated similarly in all cases when the limit of the measuring appliance is known, and the minimum temperature interval specified. For work above a red heat, the leads from the coil should always be made of platinum. Copper leads, when heated, give off vapour in sufficient quantity to attack the platinum; and the same applies to a greater degree to all kinds of solder. For low temperature work, however, copper leads may be used, thus reducing the cost of the instrument. Mica, above 1000° C., tends to crumble; and most forms melt at 1300° C. or lower; hence a mica-wound instrument should not be used continuously above 1000° C. The fireclay winding used by Siemens permits of occasional readings being taken up to 1400° C., and the same applies to wires wound on obsidian (melting point = 1550° C.), or those in which the coil is self-sustaining. As previously mentioned, however, alterations in the platinum itself render continuous readings above 1000° C. inaccurate after a short time.