]\gt ; Dr. G. W. C. Kaye .
[ May 25 , way .
Hence , as each term is positive , so that the series , cannot osciJlate , it must converge .
Moreover , we have where , by S4 , provided only .
It will be noticed that , for , this gives us information as regards the convergency of the series far inferior to that we already possess .
We have , however , no reason to suppose that , if , any theorem as striking as that for holds .
In the case in which , the above result affords us in any case a material addition to our existing knowledge .
A Silica Standard of Length .
By G. W. C. KAYE , D.Sc .
, B.A. ( Communicated by R. T. Glazebrook , F.R.S. Received May 25 , \mdash ; Read June 29 , 1911 .
) From the National Physical Laboratory .
) CONTENTS .
PAGE 1 .
Materials for tandards of LeIlgth 430 Bronze 431 Platinum 431 Platinum-Iridium 431 Nickel 432 Steel 432 Invar 432 Fused 434 2 .
Design of the Silica Metre 437 3 .
Positions of the Points of Support 439 4 .
Position of the Plane of the Reference Lines 441 5 .
Use of the Silica Standard in Air or Water 441 6 .
Platinising the End Slabs 444 7 .
Ruling the Reference Lines 445 8 .
Annealing the Silica Metre 446 1 .
Materials for Standards of Length .
The paramount qualities which must characterise a material suitable for a primary standard of length are constancy and durability .
In addition to 1911 .
] A Standard of Length .
the absence of secular variation it should exhibit no linear thermal hysteresis , i.e. no lagging of length behind temperature .
The ideal material would furthermore , from the operator 's point of view , possess a zero coefficient of expansion\mdash ; a convenience which , until recent years , has not been even approximately satisfied\mdash ; should take a polish , permit the ruling of suitable lines , and should answer all requirements as to hardness , elasticity , density , resistance to oxidation , etc. These qualifications lead us to look for a pure substance\mdash ; chemically simple\mdash ; which will combine in itself as many as possible of the important essentials .
In this connection it will be interesting to take notice of some of the materials which have been used at different times for primary line standards of length .
The Imperial Standard Yard was made of bronze in 1843 , and it has since been found by intercomparison with its official copies , also of bronze , that variations have arisen in the lengths of the different bars .
* Copper alloys are now arded as unsuitable for the purpose ; they combine a large expansion coefficient with a secular length change , due to some molecular rearrangement , which is largely dependent on the temperature treatment .
It was for this reason that a platinum-iridium copy of the yard was made in 1902 .
The original metre\mdash ; the Metre des Archives\mdash ; was constructed of plaXinum by Borda in 1795 .
Platinum enjoys the advantage of being but little affected by time or circumstance , but on the other hand it is expensive , soft , and inconveniently heavy .
When the national Prototype Metre was constructed in 1887 , the difficulty as to the softness of platinum was met by alloying it with 10 per cent. of iridium , and the objection to the weight was overcome by *In two cases the relative variation in 10 years was nearly 4 microns , i.e. about 4 parts in a million .
See 'Memorandum on a New Copy of the Imperial Standard Yard , ' H. J. Chaney , 1906 , pt .
] , p. 30 .
Dr. G. W. C. Kaye .
[ May 25 , the\ldquo ; Tresca\ldquo ; or approximately -shaped cross-section .
Elaborate care was taken in refining and in artificially ageing the material , but one cannot refrain from feeling that it must , if only to a small extent , exhibit some of the defects of an alloy .
The platinum-iridiutu alloy admits of a high polish and permits the ruling of good lines .
Its density is about two and a-half times , and its expansion coefficient about one-half , that of brass .
Young 's modulus is , about the same as for mild steel .
Of all the pure metals , ordinarily available , and reasonably expensive , nickel is probably the most suitable for a standard of length .
It is hard , does not tarnish , is about as elastic as steel , takes a high polish , and allows the of excellent lines .
A nickel standard of section at the National Physical Laboratory has shown an increase in length in five , years of only one part in 10 millions\mdash ; a remarkably good stability .
Nickel labours under the disadyantage at present that it is difficult to work .
It is not easy to roll into tapes , and it is almost impossible to obtain a nickel line standard than 1 metre .
As to steel , whatever it can offer , they are more than counterbalanced by its tendency to rust .
Steel tapes have the great advantage of lying flat and straight , ) will take a good graduation , and are not liable to kink under slight provocation .
In field work , however , where accurate temperature measurements are practically impossible , the relatively high coefficient of expansion of steel puts tapes of this material out of court for the best work .
Invar.\mdash ; The researches of M. Guillaume on the alloys of nickel and steel have , as is well known , culminated in invar , the 36 per cent. Ni alloy , which by reason of its low expansion coefficient at ordinary temperatures has received very extended application for many purposes .
Invar enjoys further does not readily rust and it will take a high polish\mdash ; desiderata for line measures and tapes .
On the other hand , it is not a permanent alloy , and it is impossible to ensure a specified coefficient of expansion , since a very slight variation in the composition is sufficient to produce considerable alteration in the expansibility .
It is in consequence doubly important in the case of a long measure such as a tape , to look out for the effects of local heterogeneity , and it is more than usually dangerous to employ an expansion coefficient derived from a short sample of the material .
In practice , invar , procurable as such , is found to have a of linear expansion coefficients of from about to , the smallness of the coefficient the criterion of the quality .
1911 .
] A Silica Standard of ength .
483 Very occasionally a negative coefficient so large as is found , and there is , of course , no difficulty in finding nickel steels approximating to invar with much lalger positive coefficients than Invar is not very easy to work ; it is somewhat soft and indifferently elastic .
Young 's modulus is about dynes/ cm.2 , i.e. about twothirds of that of steel .
A deep scratch or a bad kink is usually followed by fracture .
Invar exhibits a considerable and gradual increase of length with time , though much ( but not all ) of this may be eradicated by suitable This secular variation may be considerably accentuated by The coefficient of expansion does not seem to yary appreciably with time , nor is it affected by shocks or artificial ageing .
Bearing in mind the unstable character of invar , one would expect it to exhibit very appreciable thermal hysteresis , and , as Guillaume himself showed , this is the case .
Thermal hysteresis of length may be defined quantitatively as follows:\mdash ; be the length of a specimen which has been long maintained at It is subjected to a rise of temperature of , and is kept at that temperature long enough for equilibrium to be attained .
If , when the temperature is lowered once more to , the length after a short time is found to be , then is taken as the of the linear thermal hysteresis H. Thus may be regarded as a residual variation of the expansion coefficient .
Guillaume represents his results on the hysteresis of invar by a palabolic curve connecting residual length variation with temperature whence Thus if , the final length of an invar metre after the cycle would be about 3 microns ( metre ) shorter than the initial .
But in * The following linear expansion coefficients , deterlnined at the National Physical Laboratory , may be of interest:\mdash ; An invar metre at the National Physical Laboratory , which has a mean coefficient of expansion of , has changed its length approximately according to the parabolic relation , where is the lengthening in microns ( metre ) over a period of years reckoning from April , 1902 , when it was annealed .
Dr. G. W. C. Kaye .
S [ May 25 , this case , it must be remembered , sufficient time has to be allowed for equilibrium to be established at the higher temperature .
Now Guillaume has noticed that the higher the temperature the more rapidly does invar attain equilibrium .
" " Equilibrium is established seven times more rapidly for each rise of temperature and the case is one of a wide class of chemical So that with a cycle of temperatures of moderate range and conducted in a reasonably short time the hysteresis of invar would not attain its full value , but only some exponential fraction of it .
It will be seen that while invar has brought with it many conveniences , we need for pr\amp ; ision work a material free from its defects .
The modern tendency in the case of nickel steels is to employ the 43 per cent. which comt ) ines great stability with a moderate coefficient of expansion\mdash ; about , not unlike that of platinum .
Both secular change and thermal hysteresis are extremely small .
The 43 per cent. alloy rusts less readily than invar , and will take a high polish .
It has a Young 's modulus of about dynes/ cm.2 .
The temperature coefficient of electrical resistance is high , so that the temperature of a tape can easily be measured in this way .
FUsed Silica.\mdash ; The extremely small coefficient of expansion of fused silica would in itself commend it as a material for a standard of length .
At ordinary temperatures the linear expansion coefficient is about which with the exception of high quality invar is less than that of any other known substance .
The expansion curve of silica is shown in fig. 1 , and , as will be seen , is approximately straight from about to 1000o .
There is a length minimum at about C. , and a maximum ( accompanied by a change point ) at 1200o C. The coefficient of expansion rises to a slight maximum at about The mean linear expansion coefficient between and C. may be represented over the range to C. by ; for the range to C. * It is , for example , a matter of common observation in accurate mercury thermometry how the rate of establishment of equilibrium of a glass varies with the temperature .
To secure , for instance , the full depression of the zero after requires 20 minutes ' heating at .
Most of the changs is complete after oue minute ; but at , say , , equilibrium is not attained until after several days .
The result is that zero points taken immediately after , are all different , but , owing to the rapidity with , which glass settles down at , the point remains perfectly fixed and unaffected by previous heat treatme1lt .
The changes are slower the more complex the glass .
See Kaye , 'Phil .
Mag 1910 , p. 718 .
A Silica of Length .
435 Thus one of the qualifications of a material for a length standard\mdash ; that of a small coefficient of expansion\mdash ; is satisfied by fused silica .
As regards thermal hysteresis one would , by analogy with glass ( see footnote on p. 434 ) , expect silica to be very good by reason of its chemically simple nature , and this view was borne out by experiments carried out by Mr. L. F. Richardson at the National Physical Laboratory .
Specimens of annealed fused silica , both clear and translucent , were FIG. 1 .
obtained in the form of end-measure rods about 45 cm .
long .
They were subjected for periods of from 1 to 90 hours to various temperatures over a range of from about to C. The values for the thermal hysteresis ( as defined above ) are tabulated below .
For the sake of comparison , the figures for invar and some common thermometry glasses ( specially designed to show a small after-effect ) are added .
The glasses are in order of increasing complexity of composition .
Dr. G. W. C. Kaye .
[ May 25 , Fused , \ldquo ; , , ( 1 ) Derived from ( 2 ) Derived from ( 3 ) Derived from ) ( 4 ) Derived from ( 5 ) Jena 59 :72 per cent. , 11 per cent. , 5 per cent. , 12 per cent. Jena 16 :67 per cent. ) , 14 per cent. , 7 per cent. , 12 per cent. and Verre per cent. , 12 per cent , 14 per cent. per cent. , 2 per cent. and It will be noticed that Jena 59 has the simplest composition and the smallest value of H. Thus the hysteresis exhibited by silica was small and negative after high temperatures , and was too small to be measured after moderate temperatures .
In fact , for the temperature treatment to .
which a silica length standard is likely to be subjected , may be regarded as negligible .
It should be borne in mind that these results refer to annealed silica .
Specimens ected to a first heating frequently behave anomalously and show larger values of , which , , assumes a normal value for subsequent heatings .
Thus form a part of the history of a silica length standard .
In respect of both expansibility and hysteresis there is practically nothing to choose between the different kinds of fused silica : the cheaper airstreaked satin-like variety is as good as the more expensive clear transparent kind .
Neither kind can be regarded as costly .
Both varieties satisfy thus far the conditions we have enumerated as desirable in a material for a length standard .
Young 's modulus is about , about half that of brass ; but , on the other hand , the density of silica is low , about .
The tensile strength appears to be of the order of dynes/ cm.2 , which is a little less than half that of piano wire .
Silica is , of course , unaffected by oxidation ; its durability is undoubted .
It is a little harder than , and bout as brittle as , glass , but the care which is naturally bestowed on a primary standard would act as a safeguard against the possibility of disaster arising from fracture .
The solubility in water appears to be of the order of 1 part in 100,000 , 1911 .
] A of Length .
which , though slight , suggests that the reference lines on a silica line standard should be protected , if it is to be used in water .
We come now to the ability of silica to take good reference lines .
It can , of course , be ground and optically worked .
With a suitable diamond there is no difficulty in ruling sharp-edged lines on silica which are sufficiently fine and definite for the purpose .
Ths diamond requires to be very lightly loaded and the ruling be done extremely slowly .
But two objections appear to the of lines on the bare silica .
One is the shelling or flaking of the edges of the lines which occurs within a day or two of their ruling , and effectually prevents their use for any kind of precision measurement .
The other is the great difficulty of seeing the lines at all when the silica is immersed in water .
The refractive indices of silica and water are not very diffelent .
The rapid deterioration of the rulings might conceivably be eradicated by immediate annealing or by the use of an exceptionally hard luling diamond .
Various methods of illumination were tried to increase the visibility of the lines when immersed in water , and alternative suggestions , such as the insertion of invar plugs , were considered ; but , finally , it was decided that the plan of coating the silica with a film of plathnum and cutting the lines in the platinum offered the best prospects of success .
Such a film could not affect the expansion of the silica .
The initial experiments were encouraging , and this last difficulty removed , the construction of a metre line-standard of fused silica was decided upon by the laboratory .
2 .
Design of the Silica Metre .
The final design for the silica standard the form of a transparent silica tube a metre long , which merges at each end into a clear , transparent , horizontal slab of silica for the reference lines .
As already mentioned , these lines are to be cut in a platinum film , and , although it is possible to get a very permanent and adherent film , it was decided for greater safety to protect the platinum by a cover-slip .
If the reference lines are viewed through the cover-slip , the apparent length of the standard would be affected by any change with time of the tilt of the cover-slip .
To avoid this , the plan was adopted of platinisiug the under side of the slab , cutting the graduation completely through the film , and viewing the line from above through the slab itself .
In this way the light does not pass through the cover-slip , which is underneath the slab .
Considerations of stiffness and convenience suggested , for the body of the standard , tubing of about inch external diameter and mm. thick .
It was further decided , partly with the idea of preventing the tube VOL. LXXXV .
Dr. G. W. C. Kaye .
[ May 25 , rolling , to have silica supporting trunnions fused to the tube at one of the correct points of support , the other point to rest on a transverse roller in the usual way .
Fig. 2 is an outline drawing ( prepared from a photograph ) of the bar with its trunnions and supports .
FIG. 2 .
The under sides of the trunnions were ground to a blunt edge , the two portions of this edge being collinear to within mm. The end slabs took the form of semicircular discs , which were made of specially clear silica , free from bubbles ; both upper and lower surfaces were polished optically flat .
The upper and lower surfaces of each end slab and the lower edge of the trunnion were constructed to be parallel to within 1 in 500 ( i.e. to ) when the bar was properly supported .
Fig. 3 shows portions of the bar drawn to scale .
The over-all length of the bar is , and two holes , ( fig. 3 ) , were provided to permit the submersion of the bar , which , without these , would have floated in water .
1911 .
] A Silica Standard of Length .
It should be mentioned that the Silica Syndicate undertook the making of the standard , while Messrs. Hilger were responsible for the grinding and polishing of the trunnion and the end slabs .
3 .
Positions of Points of Support .
If the length of a bar is always measured under the same conditions of use , then so long as these conditions can be exactly specified and reproduced we need not concern ourselves with the method of support and the effect of flexure on the length of the bar .
However , a two-point support removes the uncertainty that exists as to pressure distribution when a bar not ideally straight rests on a surface not ideally flat ; it has the further advantage with an immersed bar of assisting to promote uniformity of temperature .
Now , as Airy*first recognised , it is desirable , from a practical standpoint , that the measured length should show a minimum susceptibility to accidental slight variations in the selected positions of the supports .
With this in mind , it follows , by the usual Bernoulli-Euler treatment , that for a long , uniform , homogeneous , isotropic bar on two symmetrically placed supports , the distance , , between the supports should equal where is the length of the bar .
This is , of course , a particular case of Airy 's well-known formula , where is the number of equidistant and symmetrically situated supports which divide the load equally .
has treated the question somewhat differently , and finds the most advantageous distance between the supports is given by , which differs but little from the result by Airy 's formula .
In the present since there are but two reference lines , which are situated , as near as may be , in the neutral plane of the bar , almost any convenient positions would serve for the two supports , provided these positions were adhered to .
The constant difference between the actual length of the bar and its horizontal projection does not concern us .
What we may most profitably seek is to minimise the effect on the constructed parallelism between the upper faces of the two end slabs , of any accidental inaccuracy in setting the moveable support .
The Airy positions satisfy this condition provided the reference lines are not too far removed from the neutral plane .
Our sihca metre presents the case of a uniform bar with a trunnion at one point of support and an additional weight ( that of the slab and its cover slip ) at each end .
Airy , 'Phil .
Trans 1867 , p. 17 .
Brooch , ' Trav .
et Mem. du Bur .
Intl 1890 , , p. 362 .
Dr. G. W. C. Kaye .
[ May 25 , If is the length of half the bar , the distance of each support from the centre , the distance of the additional weight from the centre , the slope of the bar at any point\mdash ; distance from the centre\mdash ; which lies between the support and the end , is given by where is Young 's modulus , I is the \ldquo ; area moment of inertia\ldquo ; of the cross-section about an axis through its C.G. and perpendicular to the plane of bending of the bar , is the weight of unit length of the bar , is the additional weight ( i.e. the excess weight over the uniform value which obtains elsewhere ) .
In our case , and ( obviously the weight of the trunnion cuts out ) , and for the change of slope at the end of the bar to be nil whence In our case is small compared with , so that From the dimeJlsions of the standard , it was found that is probably near the truth .
Taking cm .
, then cm .
, so that the weight introduces a 1 per ce1lt .
correction term into Airy 's simple formula .
Having to the unavoidable irregularity at the junction of each slab and the body of the tube , and the somewhat dubious assumption of perfect homogeneity , isotropy , and uniformity , we may take it that the round number 29 cm .
is sufficiently accurate for the purpose .
At this point then , the trunnion was constructed , and 58 cm .
from it a ring was etched round the tube as a permanent indication of the whereabouts of the other point of support .
Silica is much less susceptible than glass to the action of hydrofluoric acid , and as the result of trials , about hours ' exposure to the strong acid ( solution ) was found to be suitable .
* See , for example , Chree , ' Phil. Mag 1901 , vol. 2 , p. 698 .
1911 .
] A Silica Standard of Length . .
Position of the Plane of the Referervce Lines .
If ( fig. 4 ) is the reference line situated on the under side of the silica slab , then its apparent position is at I , where OIP is normal to the upper surface , and OI const , where is the rsfractive index .
FIG. 4 .
Since is fixed relative to the silica , I also is a fixed point in the silica and is unaffected by the direction of viewing , provided it is approximately normal .
Now the Airy points of support are so arranged that the ends of the bar are horizontal , and therefore any small error in setting will produce only vertical movements of the points situated in the neutral plane at the ends of the bar .
If therefore we arrange that I is in the neutral plane , small errors in setting , even though they cIJange the tilt of the silica surface , do not cause any alteration in the distance II ' where I ' the image of the reference line at the other end of the bar .
The idea was embodied in the design of the bar and thus the slabs are not symmetrical about the median horizontal plane of the metre , but are contrived so that when it is immersed in water the optical image of the reference lines shall lie in the neutral plane .
The efractive index of silica relative to water is about , which means that with a slab mm. thick , the is lifted about mm. 5 .
Use of the Sdica Standnrd in Air or Water .
It would obviously be a convenience if the silica metre could be used indiscriminately in air or water , so that the one standardisation would suffice .
But , for several reasons , the apparent length of the bar in air will be different from that in water unless the proper precautions are taken .
This difference can arise from the following causes:\mdash ; ( 1 ) The different relative refractive index of the silica slab with reference to air and to water affects the position of the image of the reference line ( see previous section ) , which cannot thus lie in the neutral plane of the bar in Dr. G. W. C. Kaye .
[ May 25 , both cases .
But as from silica to air is about , and from silica to water about , the displacement of the optical image in the slab ( which is mm. thick ) will only amount to mm. , so that any error which would be introduced on this score would be very small .
( 2 ) A want of parallelism between the upper surfaces of the end slabs becomes important when the surrounding medium is changed .
For , conone slab , suppose the slope of the upper surface is , measured in a vertical plane containing the long axis of the bar .
Then if the thickness of the slab is , the reference line on the under side appears to be laterally displaced by an amount \mdash ; see fig. 5\mdash ; which may be written since and will always be small .
The difference of the displacements in air and in water is , where is the refractive index from air to silica , water \ldquo ; and , then the difference of the displacementf is .
Thus , if , and , refer to the two end slabs , then length of the bar in air differs from that in water by Writing .
( A ) Now the standard of accuracy aimed at in this silica metre is 1 part in 10 millions , i.e. 1/ 10 micron .
We require then that each of the two terms in the expression for shall be less than 1 in 20 millions ; is about mm. , so that , dealing only with the first term , whence must be less than 1 in 15,000 .
1911 .
] Standard of Length .
Messrs. Hilger were instructed to secure a parallelism of at least between the upper surfaces , i.e. 1 part in 25,000 , the bar being properly supported .
( 3 ) The method of setting up a bar for observation in a comparator ensures that the plane which contains both reference lines shall be horizontal ; but , unless the surfaces of the two end slabs are coplanar , we shall have a state of things shown in fig. 6 , in which both surfaces , although parallel , have to be tilted to secure the horizontality of the plane containing the reference lines Thus , although vanishes , may be appreciable , so that in the expression ( A ) above for , the term may become of importance , and should be provided for in the specification .
Consider , as an extreme case , when ( in fig. ) mm. , then Now the order of the lirst term in expression ( A ) is mm. ; the order of the second term should be no more than this .
Thus } , i.e. } mm. When the present bar was ordered , the importance of this condition had not been realised , and the length of the standard in water was subsequently found to exceed the length in air by micron .
In a later specification the following stipulations have been incorporated:\mdash ; ( a ) The upper surfaces of the two end slabs be coplanar to less than a millimetre over the length of the bar , i.e. shall be less than 1 mm. ( b ) The two end slabs shall have the thickness to within mm. Obviously , this clause also provides for the unlikely case of the bar , as a whole , being used with an appreciable inclination to the horizontal .
Apart , however , from the convenience of being able to use the bar indiscriminately in air or water , there is little to urge in favour of the above somewhat stringent restrictions as ards the parallelism , thickness , etc. , of the end slabs .
At the National Physical Laboratory it is probable that the bar will almost always be used in water .
We have , moreover , to remember that the annealing of the bar may have some slight effect on , at any rate , the parallelism of the slabs .
Dr. G. W. C. Kaye .
[ May 6 .
Platinising the End Slabs .
As mentioned above , the under side of each slab is to be platinised , and accordingly a method of platinising was sought , which would give a bright surface on the side next to the silica , so that the lines , when viewed by reflected light in the usual way , would appear dark on a bright ground .
The method of cathodic sputtering in a vacuum tube can be used for producing metallic films of extreme thinness .
Thick films are apt to peel unless very slowly deposited .
The chief conditions for success are : ( 1 ) pressure maintained low enough to yield a cathode dark space of , say , about 2 inches , just within which the body to be coated should be placed , and ( 2 ) a not too powerful discharge .
A disc cathode of platinum is convenient , and it is beneficia ] to put the body ( previously cleansed ) in metallic contact with the anode .
A hydrogen atmosphere is advantageous , though usually enough hydrogen is expelled from the cathode during the discharge .
The method has the advantage that it can be used at ordinary temperatures , but the films deposited in this way , though bright , are not strongly adhering , and will not withstand any sort of abrasion .
* Recourse was had therefore to the " " liquid platinum\ldquo ; of Messrs. Johnson , Matthey \amp ; Co. This is a slightly gummy solution of a platinum salt , which , on nition , is capable of yielding excellent deposits of platinum .
These are ordinarily brightest on the side remote from the silica , but as the result of many trials it was found that , with the following precautions , a suitably bright , closely adhering film of platinum could be obtained with certainty .
The silica slab is cleaned to free it from grease , etc. , and the solution is genel'ously applied with a small camel-hair brush to the surface , which is then put on the slant and allowed to drain , the accumulated excess being removed after a few minutes .
A clear transparent film of the " " liquid platinum\ldquo ; is thus deposited evenly over the surface , and is allowed to dry slowly at room temperature for a day or so .
It is finally dried off on an air or water bath .
The silica ] is then inserted in a muffle furnaoe at a dull red heat ( 600 C. ) and allowed to remain there for a few minutes .
On withdrawal it will be found to be coated with a film of platinum which is almost equally brilliant on the two sides , but is , however , soft and readily detachable .
The furnace temperature is raised to about 90 ( bright cherry red ) , and the end of the bar re-inserted for two or three minutes ; a longer time is apt to spoil the lustre .
This treatment renders the platinum-film hard and strongly adhering ; it would , if necessary , stand burnishing with some soft *Sir David Gill informs me that sputtered iridium films can be got hard and adherent .
1911 .
] A Silica Standard of Length .
445 metal .
The films thus prepared are semi-transparent , and should , when held up to the light , present a homogeneous and fine-grained structure .
If desired , a second film can in like manner be superimposed on the first .
The method is unobjectionable on the score of the safety of the silica , which can be subjected to the sudden changes of temperature with impunity .
The necessity for employing such a high temperature in platinising made it desirable to carry out the annealing of the bar at a stage subsequent to the platinising , and this procedure was adopted .
7 .
Ruling the Referencc Lines .
It was decided to rule the reference lines before annealing , in order to ascertain any effect that the annealing might produce on the of the bar .
The lines were ruled by the laboratory dividing engine .
Apart from the choice of a suitable diamond , the conditions for success are a light load and slow ruling .
The comparatively soft platinum deposit , backed up by the harder silica , permits the ruling of very beautiful clean-edged lines .
Almost any degree of fineness can be obtained by varying the load on the diamond , the fineness depending , of course , on the extent to which the diamond removes the platinum film .
The method of observing the lines through the silica has the great advantage that the narrowest part of the furrow cut by the diamond is the only part that can be viewed by the microscope .
FIG. 7 .
FIG. 8 .
The lines lend themselves to work of the highest precision , and constitute a valuable addition to the advantages which the silica metre possesses .
Comparison with the lines on the nickel and invar standards of the laboratory gives results entirely in favour of the lines on the silica standard .
For loads less than about 5 grm. , the lines were inconveniently fine for use in comparator microscopes of 20 or 50 diameters nification : a load of 7 .
was finally adopted .
A couple of transverse\ldquo ; finder\ldquo ; lines , each 100 microns from the central 446 Dr. G. W. C. Kaye .
[ May 25 , transverse reference line , were inserted on each end slab .
They were crossed at right angles by a pair of longitudinal lines about 200 microns apart .
The part of the central transverse line which is contained within the two tudinal lines is that which is used in measurement ( fig. 8 ) .
The sihca metre was then compared with an invar metre , and its length was found to be metre at C. 8 .
Annealing the A reference to the expansion curve of fused silica on p. 435 will show that at about a not very pronounced maximum value of the coefficient of expansion is indicated .
Quartz crystal at about the same temperature exhibits abnormally high expansion , and perhaps it would be right to infer that the maximum at 50 in the case of fused silica is a residual effect due to traces of unconverted quartz .
However , the point , such as it is , is useful in setting a limit to the temperature that should be employed in the silica metre .
Moreover , the table on p. 436 shows that the thermal hysteresis exhibited by silica is much less for temperatures below about 50 than for higher temperatures .
In the absence of any guiding information , it was decided to go no higher than about in the annealing .
Experience and future observation of the standard may indicate whether or not a lower temperature would be more advantageous .
For the purposes of the annealing , an electric furnace of a fireclay tube wound with Eureka wire was constructed .
The length of the tube was about 6 feet , its external diameter about inches .
The lagging was of magnesia brick and very ampJe .
An exploring Pt-Ir thermocouple was used to calibrate the furnace , and showed that the temperature was approximately uniform hout the central region employed .
A special fireclay carriage with silica contacts for the correct supports of the metre was built , and the metre was inserted into the furnace at room temperature .
By suitable external resistance control , the temperature was raised in two days to about 45 C. This temperature was maintained for some eight days , after which about a fortnight was taken for the furnace to cool down slowly to room temperature .
The length of the metre after annealing was found to be metre at C. , which points to a shrinkage of a little over half a micron due to the annealing .
This is extremely small after such severe temperature treatment , andgives support to the view that the future secular variation of the silica standard will be very slight .
With this point in view , is intended to keep the silica metre under close observation during the next few years .
Semicircular silica cover-slips were affixed under the end slabs by means of 1911 .
A Silica Standard of Length .
447 a cold cement kindly supplied by Mr. Chas .
Baker .
The joints were afterwards made water-tight at the edges by a mixtule of red lead and linseed oil .
9 .
Summary .
The general properties of fused silica , and iu particular its remarkably low coefficient of expansion , render this substance specially suitable for the construction of permanent length-standards of the highest class .
The coefficient of expansion of platinum-iridium , which has hitherto been the material almost exclusively employed in the best work , is about per degree C. , while that of silica over the ordinary range is about , i.e. , 1/ 20 of this amount .
It is true that the best qualities of invar , M. Guillaume 's nickel-steel containing 36 per cent. Ni , have expansion coefficients comparable with that of silica , but experience has shown that while invar is eminently useful for working standards , it is quite unsuitable for primary standards , owing to its large thermal hysteresis .
Fused silica , on the contrary , has now been found to be practically entirely free from this defect ; it enjoys , in the matter of cost , an enormous advantage over platinum-iridium ; furthermore , in view of the fact that primary standards are always handled by trained and skilled observers , its comparative fragility is of small consequence .
Modern methods of manufacturing and working silica have rendered it possible to construct a silica line-standard metre .
The present model , the first of its kind , consists of a silica tube into which are fused at its ends optically-worked plane parallel slabs of silica .
These carry the graduations , and their undersides are platinised .
The graduations defining the metre length are made by cutting through the platinum film with a ruling diamond .
The platinum deposit permits the ruling of very beautiful cleanedged lines .
The bar is supported at the Airy points and so that the slabs are horizontal .
The lines are viewed through the slabs from above , and are thus seen to advantage .
The apparent length of the standard is independent of any change of tilt of the cover-slips which are used to protect the platinum films .
The thickness and position of the end slabs are so arranged that the image of each reference line lies in the " " neutral plane\ldquo ; when the bar is immersed in water .
The writer wishes gratefully to acknowledge the continual kindly interest of Dr. R. T. , Director of the National Physical Laboratory .
He would further record his thanks to Mr. J. E. Sears for very valuable criticism and advice at many stages in the work .

