402 schlink: variance of measuring instruments 



performance with regard to variance, of instruments of diverse 

 design or workmanship. 



Far from being a relatively unimportant source of inaccuracy 

 in measuring instruments, it can be shown that the hysteresis 

 or variance type of error demands consideration in practically 

 every type of instrument, while in some (such as pointer-and- 

 dial types of displacement-indicators) the variance is a prepon- 

 derant factor in design and actually sets the limit of sensitivity 

 and accuracy practicably obtainable, so limiting the application 

 or utility of the instrument. 



The phenomenon of drift, which is a time effect characterized 

 by more or less gradual movement of the indicating element 

 asymptotically to a definite reading, after all conditions external 

 to the instrument have become constant, is one requiring investi- 

 gation from the point of view of the elastician, and although in 

 the case of some instruments it is an important cause of vari- 

 ance, it cannot be adequately treated in the present paper. 



Reducing variance in the mechanism. A few effective means of 

 reducing instrumental variance are readily available. The 

 conical pivot may advantageously be substituted for the cylin- 

 drical journal and bearing, thus affording the type of bearing 

 familiar in the balance wheels of alarm clocks and low priced 

 watches. Such a bearing has a low frictional moment, and main- 

 tains a practically invariable relation of bearing to journal but 

 cannot, of course, support any considerable load. The use of 

 flexible or ribbon-like connectors secured by simple clamps at 

 their extremities, between members which are to be conjoined, 

 instead of the more usual rigid pin-and-link connectors is ofttimes 

 a very useful device and has the advantage of providing for 

 correction of motion to obtain a linear scale of graduation, by 

 the simple expedient of arranging that one or both ends of the 

 tape shall wind upon a cam of suitable contour. This type of 

 connector contributes very little indeed to the hysteresis of the 

 complete mechanism, since practically the only source of irre- 

 versibility in such a tape is the hysteresis of inelasticity, which in 

 absolute value is very small. Simplifications of mechanism are 

 often possible to eliminate the number of links involved and 



