TY - JOUR
T1 - Experimental evaluation of instrumentalcale readings obtained by eye interpolation of fractional interval values
AU - Tomita, Yutaka
AU - Honda, Satoshi
PY - 1994/4
Y1 - 1994/4
N2 - This paper describes an experimental evaluation of the precision achieved by eye interpolation. It explains the effect on eye interpolation of differences in the position of a pointer. Fiducial lines-a fiducial line is the "0" or leftmost line on a vernier scale and serves as a coarse pointer-were set at different positions between 0 and 1 mm (0.05 mm intervals) using the vernier scales. The vernier scales were then covered with plastic tape. Thi vernier callipers were randomly given to 19 whbjects who were asked to read the position of the fiducial lines on the main scale, and to estimate them to the nearest tenth of a division. As a result of this experiment we found the following: (i) The bias and dispersion of readings just below the scale marks were the smallest. (ii) When the fiducial line were located at less than 0.5, the readings were biased toward 0, and when they were located at greater than 0.5, they were biased toward 1. (iii) Based on an analysis of variance involving the position of the fiducial line, the individual observers, and their interactions, the position of the fiducial line was found to be the most significant. (iv) Thus, personal errors in reading should be categorized together with instrumental errors, and could be reduced with maximum likelihood estimation.
AB - This paper describes an experimental evaluation of the precision achieved by eye interpolation. It explains the effect on eye interpolation of differences in the position of a pointer. Fiducial lines-a fiducial line is the "0" or leftmost line on a vernier scale and serves as a coarse pointer-were set at different positions between 0 and 1 mm (0.05 mm intervals) using the vernier scales. The vernier scales were then covered with plastic tape. Thi vernier callipers were randomly given to 19 whbjects who were asked to read the position of the fiducial lines on the main scale, and to estimate them to the nearest tenth of a division. As a result of this experiment we found the following: (i) The bias and dispersion of readings just below the scale marks were the smallest. (ii) When the fiducial line were located at less than 0.5, the readings were biased toward 0, and when they were located at greater than 0.5, they were biased toward 1. (iii) Based on an analysis of variance involving the position of the fiducial line, the individual observers, and their interactions, the position of the fiducial line was found to be the most significant. (iv) Thus, personal errors in reading should be categorized together with instrumental errors, and could be reduced with maximum likelihood estimation.
KW - Analysis of variance
KW - Calibration with maximum likelihood
KW - Personal error
KW - Space eye interpolation
UR - http://www.scopus.com/inward/record.url?scp=0028407774&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=0028407774&partnerID=8YFLogxK
U2 - 10.1016/0263-2241(94)90007-8
DO - 10.1016/0263-2241(94)90007-8
M3 - Article
AN - SCOPUS:0028407774
VL - 13
SP - 147
EP - 151
JO - Measurement
JF - Measurement
SN - 0263-2241
IS - 2
ER -