All materials change in size to a greater or less degree with variation of temperature. It is there fore essential to specify exactly the temperature at which the material standard defines the unit of length, and further either to control the standard exactly to this temperature when making comparisons with it, or at least to ascertain its temperature exactly, and make allowance for its known expansion between that tem perature and the standard temperature. The Imperial Standard Yard is correct at 62° F, the International Prototype Metre at o° Centigrade. The numerical relationship between the two units has been twice accurately determined : in 1895 by Chaney and Benoit, with the result 1 metre =39.370'13 in.; and more recently (1922– 25) at the National Physical Laboratory, with the result 1 metre= 39.370147 inches. The two results may be said to be in agree ment within the experimental error of the various observations in volved, and for all practical purposes the simple ratio i in.=
mm. (equivalent to 1 metre =39.37°079 in.) is exact.
The use of this ratio implies that the two objects being com pared are both simultaneously at their common temperature of employment. For all everyday purposes of measurement, as for example in measuring a piece of work in an engineering work shop, it is unnecessary to bring the object to be measured to the standard temperature. If the work and the gauge with which it is being measured are made of materials having the same co efficient of expansion, and the former has been compared with the reference standard at the standard temperature, it is only neces sary, when comparing the work with the gauge, to ensure that both are at the same temperature (not necessarily the standard tempera ture) to ensure that the work will be correctly measured.
For many purposes where very precise meas urement is involved, it is of great advantage to have a material with a very small thermal expansion. Two such materials have been discovered. The first, known as "Invar," is a nickel-steel alloy, containing 36% of nickel, invented by Dr. Ch. Ed. Guillaume of the Bureau International (Sevres). Different samples have co efficients ranging from about 1.5X
per I° C, for large bars, down to zero, or even slightly negative values, for smaller bars, pendulum rods, drawn wire or rolled tape. These figures are to be compared with 'IX i
per I° C for steel, and i8 X
per I° C for brass. Invar, unfortunately, has one very serious defect as a standard of length. It grows longer, rapidly at first and subse quently more slowly but continuously, so that after 20 years the length of a bar of invar will still be increasing at a rate of about I part in 2,000,000 per annum. At a later date Dr. Guillaume
introduced a slightly different alloy, containing a percentage of chromium in addition to the nickel, which is described as "stable" invar. This alloy grows at an appreciably slower rate than ordi nary invar, but still cannot be regarded as constant. Invar there fore is principally useful in a laboratory, in work where temporary constancy of length is the primary consideration, or in circum stances, as for example in the case of tapes and wires used for geodetic surveying, where the accurate ascertainment of tempera ture presents considerable difficulty. In these cases it is necessary to return the tapes or wires to the laboratory periodically for reverification. For the reasons indicated above, in everyday work shop measurements, gauges or scales should be made of material having a similar coefficient of expansion to the work to be meas ured, and invar therefore should not ordinarily be used for the construction of workshop standards.
The other material which has a very low coefficient of expansion is fused sil ica, which expands only 0•4X
per I° Centigrade. A metre standard constructed of fused silica, in the form of a tube, with parallel plates fused in at the ends on the platinised surfaces of which the defining lines are ruled, has been made and kept under observation at the National Physical Laboratory since 1913. So far as can be detected, no change has occurred in its length. Such a standard is extremely fragile, and for this reason would hardly be adopted either as a fundamental reference standard or for everyday use, except in a metrological laboratory.
Reference should be made to the recent work of Perard, at the Bureau International, on end standards of natural crystal quartz. Such standards cannot, of course, be of the full length of the yard or metre. But they present several great advantages. Firstly, since the material is of great age and the molecules of which it is composed are arranged structurally in definite crystalline array, there seems little possibility of any secular change. Secondly, it lends itself to perfect optical finish of the defining end planes, which enables direct determination of length to be made in terms of light waves, with extremely high precision resulting in a propor tional accuracy no less than is at present obtained in the compari son of yard or metre line standards, although the largest available specimens of crystal quartz, as well as the method of use, only enable such standards to be verified up to about 4 inches.