385 A Standard Measuring Machine .
By P. E. Shaav , B.A. , D.Sc .
( Communicated by Prof. J. H. Pointing , F.R.S. Received June 4 , 1912 .
) [ Plate 11 .
] In some physical researches it is required to measure the dimensions of a regular solid , whether parallel-sided , cylindrical , or spherical-ended , with the greatest possible accuracy .
Again , in the regular work of a metrology bureau , accurate comparison of the size of an end-standard of length with reference to a line-standard is required .
Until 1906 the only apparatus available for these purposes was the measuring machine made for comparing engineering gauges .
In that year the writer described a machine* based on the principle of electric touch , which was much more delicate than the older mechanical machines .
It also had facilities for exploring the solid under test , to prove its accuracy of figure .
This machine has been installed and used in the National Physical Laboratory since 1909 .
Improvements in it were described later.f The present paper gives an outline of an improved machine of the same type embodying the experience gained in using the 1906 machine .
The chief novelties are : ( 1 ) greater strength in the supporting parts , and consequent rigidity of the whole apparatus ; ( 2 ) larger and much improved table to carry the solid under test ; ( 3 ) improved measuring-ends ; ( 4 ) an innovation for making absolute measurements , whereby the line-standard moves but the reading microscope remains fixed ; ( 5 ) side girders to relieve the bed of load and so reduce friction , abrasion , and strain .
The engineering machine serves well where quickness rather than extreme accuracy is required .
The present machine is constructed so as to avoid the following defects in the engineering machine : ( 1 ) lack of adjustments , the observer not having the means of testing and " trueing " the mechanism ; ( 2 ) assumption of the truth of the measuring-ends and of certain other flat surfaces employed ; ( 3 ) no provision for setting the solids to be measured accurately along the line of measurement ; ( 4 ) application of considerable forces during measurement .
The accompanying plate shows the new machine ( fig. 1 ) as set up in the works .
In use in the laboratory the machine rests on a concrete table , built from the foundation , and there are certain fittings not shown in the VOL. lxxxvii.\#151 ; A. * ' Boy .
Soc. Proc. , ' 1906 , A , vol. 77 .
t ' Boy .
Soc. Proc. , ' 1910 , A , vol. 84 .
Dr. P. E. Shaw .
[ June 4 , plate : { a ) The electrical circuit and telephone ; ( b ) the fixed microscope used for reading the scale D , and rigidly attached to the concrete table .
The complete length over all is 8 feet , and total weight about 800 lbs. At the extreme front and back are seen parallel girders mounted on pillars .
These carry the chief weight of the movable parts AD , B , C , the load being conveyed through vertical spring boxes to the ball-bearing castors running on the girders .
The girders serve several purposes : ( 1 ) to relieve the bed of most of the load ( about 200 lbs. ) of the movables , and so reduce flexure of the bed ; ( 2 ) to render the movement of the movables on the bed easy ; ( 3 ) to reduce the wear on the bed by lessening the pressure on the bed of the five feet of each movable .
Inside , but not touching the girders , is the bed of the machine , resting on the concrete bed by three feet .
There are two bow-shaped vertical webs forming the sides , 15 inches deep at the centre , tapering to 6 inches at the ends .
These are 1 inch thick , increased to 2 inches at the edge flanges .
The webs are joined by nine vertical cross webs , also 1 inch thick .
One main web is surmounted by a horizontal plane surface , and the other web by two plane surfaces inclined at 90 ' to one another , and each at 45 ' to the horizontal .
Each web is pierced by six holes to reduce dead weight .
It will be seen that this bed is unusually strong .
The bed was cast and aged with great care , the process of repeated scaling , ageing and final scraping taking nine months .
Four months after the final scraping the writer applied his tester for plane surfaces* and found the cumulative error from truth of the bed surfaces from end to end was not more than 0*01 mm. , which is negligible .
The central movable part , or table ( B in the plate ) , carries the solid under test and provides it with the rotations and translations necessary .
As explained in the former paper , f two rotations are required to set a parallelfaced solid true in the line of measurement , and two translations are necessary for exploration of the faces .
The way in which these are obtained ( similar in general principle to that shown in the two former papers ) cannot be shown without detailed drawings .
The greatest care was taken in every detail of design and manufacture , as practice has shown that the table is more liable to error than any other part of the mechanism .
The proof of efficiency lies in the accurate repetition of readings ( see tables ) . .
The headstocks A and C carry micrometer screws .
A is attached to a long bracket carrying the line-standard D , whereas C is plain .
These * 'Boy .
Soc. Proc. , ' 1912 , A , vol. 86 .
t Log , cit. A Standard Measuring Machine .
1912 .
] micrometers read to OT micron ( = 1/ 10,000 mm. ) , and will be understood from the former paper ; but fig. 2 shows two additional details : a , the reading-microscope for the divided circle of the micrometer ; 6 , the measuring end .
The line-standard has its ruled surface carefully set level with the measuring-ends .
It is obvious that it is sounder to have a fixed microscope and moving scale , as here arranged , than a fixed scale with microscope movable , as is usual .
In the latter case , any error in that part of the bed traversed by the headstock produces direct error in the reading , whereas , on the present plan , no error of the first order can arise , if the bed surfaces be even moderately true .
The present complex form of the measuring-end , the details of which cannot be given in the present outline , is designed so that : ( 1 ) The electrical contact shall be made , not on the solid under test , but between two smooth , clean , surfaces of iridio-platinum .
( 2 ) A flat end-surface which can be accurately adjusted true .
The measuring end has many advantages over the old form , viz. : ( 1 ) Non-metallic solids can be measured .
( 2 ) The adjustable flat ends are convenient for many measurements , but for exploring the ends points are generally used .
( 3 ) Contact sounds made between two surfaces of iridio-platinum are clearer than between one of platinum and one of iron , brass , etc. ( 4 ) The cleaning of the contact surfaces is easy and causes little thermal expansion , invar being used through most of the projecting end .
Before commencing measurement the writer adjusted the upright reference plane of the table true , and set the run of both micrometer screws parallel to that of the bed .
The following are the possible errors arising from the use of the machine :\#151 ; ( 1 ) Bed irregularities\#151 ; Flexure and pit-marks together produce negligible error as shown by the surface-tester ( see above ) .
( 2 ) Irregularity in the upright plane of the table\#151 ; The greatest error observed in a set of observations on this point was 0*0015 mm. ; the resulting error , being a cosine one , is negligible .
( 3 ) Temperature change\#151 ; The bodies , if long , are swTaddled in flannel to minimise temperature change , which is slight in a thermostatic room , and a thermometer bulb is placed between the flannel and the solid .
A change of 0*1 ' C. in a 4-inch iron bar is equivalent to one unit in the micrometer .
( 4 ) Calibration errors of the screws\#151 ; This was worked out in the former paper .
( 5 ) Microscope readings\#151 ; Uncertainty in reading the line standard is about 0*3 micron for a single reading with the invar standard , H pattern , used .
( 6 ) The irregularity of the flat end-surface\#151 ; These are circular , \ inch in diameter , and the departure from true fiat as seen by interference rings is 0*2 micron , and as the end-surfaces are both slightly convex , the resulting error on measurements is negligible .
Dr. P. E. Shaw .
[ June 4 , An electric circuit containing a cell , resistance of 5000 ohms , telephone and switch , is connected to the movables , A , B , C. The switch enables circuit to be completed in three ways , A to B , B to C , A to C , when the contact surfaces in the measuring-ends are touching .
A and C are insulated from the bed .
As an example of the working of the machine , suppose it is desired to compare a simple 1 inch Johansson gauge ( A ) with a composite 1 inch Johansson gauge B ( 0*40 + 0*35 + 0*25 ) .
Insert these side by side in the table , clamp their centres and wrap in flannel and leave to settle .
Use the point end-surface .
First set the left side of A by making electric contact and working the adjusting screws till contact readings all over the gauge surface agree ( see former paper for details ) .
Let the rectangular figure represent the left face of the gauge .
Mark off 11 places regularly spaced as shown ; enter them in column 1 , Table I , translate the gauge till the contact end is at place 1 .
Take the left micrometer reading ( 157*3 ) and enter it in column 2 .
Change the switch and bring up the right contact point to the right face of the gauge .
Take the micrometer reading ( 352*4 ) and enter it at top of column 3 .
Add the two readings and enter the result ( 509*7 ) at top of column 4 .
Proceed thus for all 11 places on the gauge and obtain the resultant column 4 .
The numbers in column 4 do not signify actual length of the gauge , but the difference between the numbers shows how the gauge differs in length according to the line used for measuring .
Thus the greatest difference in length given by column 4 is 0*4 micron .
Repeat the whole process four times , obtaining resultant columns 7 , 10 , 13 .
The arithmetic means are in column 14 .
The whole process of filling this table will occupy , perhaps , one and a half hours , and , if there is no progressive change to indicate alteration in temperature , the results in column 14 may be accepted as showing accurately differences all over the gauge .
Remembering that high reading shows smaller gauge length than low reading , it is seen that this gauge is slightly wedge-shaped , the left side being about 0*2 micron larger than the right .
Having completed the test of gauge A , proceed to set the composite gauge B. Take measurements as before and so obtain Table II .
The resultant column 11 shows that the bottom end of the gauge is smaller than the top , whereas right and left are nearly identical .
From the Tables I and II , it is evident that errors due to the machine average about 0*13 micron , whereas the gauge errors are distinctly greater than this amount .
It is easy , provided the gauge remains fixed , to repeat 4 11 7 3 10 6 2 9 5 1 8 Table I.\#151 ; 1-inch Johanss 1912 .
] A Standard Measuring Machine .
14 .
Means .
H CO^O T ?
rf\lt ; H. Q G5 cj 99999999999 05C505C505050C5000 O O O O O O rH O pH rH rH VOVOVOVOVQVOVOVOOVOVO , f4 CO + 509 *9 509 *85 509 *7 509 *95 509 -95 510 0 510 *0 510 *05 510 *0 510 *05 510 *1 9\#171 ; VO vO VO VO 9COHHM999HO9 COCO\lt ; MCOVOCO\lt ; MH^HVO VOVOVOVOVOVOVOVOVOVOVO COCOCOCOCOCOCOCOCOCOCO VO VO VO pvpcpfN^ocoqcqoiw^ COCOi^CD^COlHVOVOvOrJi VOVOVOVOVOVOVOVOVOVOVO HHrtrlHHrlHrtHH .ri s + Hi i\#187 ; 999999 ?
997i 0505050505C50J5000 OOOOOOHOHr-lrH VOVOVOVOVOVOVOVOVOVOVO pvppp^i\gt ; T ?
xvpoqvp ( Moqcqcq^fMoqcococovo VO VO VO VO VO VO VO l O VO VO VO cococococococococococo 0014 9 tp 9 9 9 9 9 h 0 9 9 CO t\gt ; t\gt ; X\gt ; VO t\gt ; CO CO CD ^ VOVOVOVOVOVOVOVOVOVOVO H rH rH H rH rH rH rH r\#151 ; i rH rH .\#171 ; *\gt ; + Hi 99999999O9O 0505050505 OOOOOO OOOOOrHrHrHrHi\#151 ; 1 H vovovovovovovovovovovo VO VO VO 9^^99999^9^ ( ?
q\lt ; M\lt ; M(Mvoco\lt ; MvoTjicqvo vovovovovovovovovovovo cococococococococococo VO VO VO VO p^rHTffpvHMpvpOCp G5i\gt ; .t\gt ; ^^CO^VOVOQ0^ lOVOvOiOVOvOVOvOvOVOVO HHHHHHHHHHH .rf + h !
^*999999971719 050505000000005 O O rH r-H rH rH rH rH *-H O lOiOlOlOlOiQiOiOlOiOlO copj ^999991-99^9 NCOCO-HOH(COvOHt'HVO vovovovovovovovovovovo cococococococococococo 9999999^h99 ncocdvo-^cded^cocoh* vovovovovovovovovovovo rH rH rH rH H pH rH iH rH rH rH 1 .
Place .
HIMCO^VOCO^OODOH rH rH | id a a .
rH .
r\#151 ; I .pH o o o OlOlO rti co cq o o o \/ a o CQ m cs rG O o .
rH hH I\#151 ; I c3 H 03 rH OHQOOOQQOpaS .^TfCOCO^COCO^^COCC cococococococococococo ooooooooaoaoooooooocuo iNOQOoSftHOOJN VO \lt ; bt\gt ; vovococ5cocoi\gt ; vovo VO VO VO \#169 ; TfOWippQOHOOON I\gt ; COC500t^TFi\gt ; 00.t\gt ; 0000 VO^OVOIOIOVOVOVOVOVOVO cococococococococococo \#171 ; J\gt ; + H VO VO VO 995999900999^ HC0C0HlH\lt ; C0C0H\lt ; '^ , C0C0 cococococococococococo ooooaoaoaoooooooaoooao \lt ; 6# VO VO p^vpvp^rort\lt ; aoa)i\gt ; ^ H 00 05 H CO CO N l\gt ; QO CO CO ^vp^^oocp^oqoqoico ( MO^\lt ; MOOHCOi0 1\gt ; N COCOCO^OCOCOCOVOVOVOVO COCOCOCOCOCOCOCOCOCOCO ^ + Hi VO VO VO VO VO H9999999999 TjH^TflCOTfi^CO^TftCOCO cococococococococococo 00 00 00 00 00 00 00 00 00 00 00 cop^ ' pvopH^iHJHcq^oop VO VO VO VO ONf NOOMN9999 O5NOO5OO0HOOOHH JOVOCDICVCVCCD0VOCO\#169 ; cococococococococococo \lt ; 35 H S s + Readings are expressed in microns ( 1/ 1000 mm. ) , A Standard Measuring Machine .
readings to 0 05 micron ; this shows that right and left micrometer are free of backlash and like errors , but in this exploration work the gauge has to be moved parallel to itself through a cycle , and it has been a difficult mechanical problem to provide such true parallel movement that readings at any one place on the gauge will agree to the accuracy of Tables I and II .
Probably the table could yet be improved so that the mean errors would be less than OT micron after a cycle .
A great variety of gauges and other solids , parallel-faced , cylindrical , and spherical-ended , of English and foreign make , have been examined by this machine .
The Johansson bars are the best worth testing , other makes being in general of a lower order as regards accuracy of figure .
No previous machine of this or other type has been accurate enough to prove errors in the Johansson gauges , but with this machine , if care be taken , it is possible to detect errors of the order shown in the tables . .
As to absolute measurements of length , these are best made with the flat surface-ends shown in fig. 2 ( Plate 11 ) .
But these ends must be " trued " as follows : A contact point is clamped in the table and so can be provided with two translatory movements .
If the surface-end is true the micrometer readings when contacts are made at various places on it will agree .
If not , agreement is obtained by three adjusting screws .
Procedure for absolute measurement is then straightforward and will be understood from the description given above .
The above paper gives , in outline , the scope of the machine .
Full details and tables of results will be published later .
The machine took 18 months to construct , and when completed was set up in the metrology department , National Physical Laboratory .
The writer 's thanks are due to Dr. E. T. G-lazebrook for encouragement and for facilities afforded in the long work of installation , and to Dr. T. E. Stanton for much help in obtaining skilled mechanical assistance for alterations during the testing of the machine .

