INSTRUMENTAL variance, which forms the major part of the discussion of this paper, is a factor whose effect in causing error in measuring instruments has hitherto been given but little consideration. In order to bring out the esser~tial relations of variance characteristic, the paper opens with a discussion of accuracy and sensitivity, showing how each of these terms can be given precise definition with definite numerical significance. Sutch values when detennined have important use in establishing the figure of merit or value of a given instrument or type of instrument. A distinction is dramn between the commercial accuracy commonly called for in direct reading instruments of the shop and plant, and the higher grade of accuracy required in the laboratory where commercial indications are normally subject to ~he application of corrections for'error .of calibration.. The error arising from whatever source, observed in an indication o.f the instrument, divided by the true value of the measured quantity, may be ,termed lJhe inaccuracy a,t a given reading, the negative :term being justifiable on the basis of custom and ease in appli,cation. The accuracy will be the reciprocal of the quantity just defined. The observed error of an instrtunem (or .~he correction) is utilized for adjusting observations before their use in calculation, while the inaccuracy and accuracy find their utilization in rating :the quality o.r performance of the . instrument.
An instrument is understood to be sensitive when it shows a change of reading for any change in the quantity being measure.d, a definition having little significance until means are available for expressing the sensitiveness numerically. Attention is called to the common inconsistencies in the sensitiveness rat-* Communicated by the Director.