NUMBERS STATED IN THE APPENDIX ΤΟ THE REPORT.

IN SUCH AN INTERCOMPARISON, ONLY DIFFERENCES IN LENGTH CAN BE MEASURED. BECAUSE OF THE LIMITED RANGE OF PRESENT COMPARATORS, ALL ARTIFACTS IN A GIVEN COMPARISON ARE OF THE SAME NOMINAL VALUE. A REDUNDANCY IN THE NUMBER OF MEASUREMENTS (EIGHT MEASUREMENTS TO DETERMINE FOUR VALUES) PROVIDES A MEANS FOR CHECKING ON THE PRECISION OF THE PROCESS BY THE METHOD OF LEAST SQUARES USING THE SUM OF THE LENGTHS OF THE TWO REFERENCE ARTIFACTS, ((.) + (..)), AS THE RESTRAINT. THE COMPUTED DIFFERENCE BETWEEN THE REFERENCE ARTIFACTS ((.) - (..)) IS AN INDEPENDENT ESTIMATE OF THE DIFFERENCE. AND SERVES AS A CHECK STANDARD'.

ALL

MEASUREMENTS FOR THIS REPORT
WERE MADE IN A TEMPERATURE
ENVIRONMENT IN THE NEIGHBORHOOD OF
20 C. ASSIGNED VALUES HAVE BEEN
ADJUSTED TO THAT APPROPRIATE FOR
USE IN AN ENVIRONMENT OF 20 C
(1968 INTERNATIONAL PRACTICAL
TEMPERATURE SCALE) USING THE
STATED. OR HANDBOOK, VALUES FOR THE
TEMPERATURE COEFFICIENT OF LINEAR
EXPANSION. IN THE COMPARISON
PROCESS. ALL ARTIFACTS ARE AT VERY
NEARLY THE SAME TEMPERATURE.
CORRECTIONS BASED ON DIFFERENTIAL
COEFFICIENTS OF EXPANSION ARE
ASSUMED NEGLIGIBLE FOR ARTIFACTS
OF GROUPS I THROUGH VI. IN ORDER
TO EXTEND THE USEFULNESS OF THE
ASSIGNED VALUES OVER A TEMPERATURE
RANGE OF 20 C TO 25 C, IT MAY BE
NECESSARY TO DETERMINE A
COEFFICIENT OF EXPANSION FOR EACH
ARTIFACT OF GROUPS VI AND VII. A
PROCEDURE TO DO THIS IN THE NORMAL
COURSE OF MEASUREMENT IS NOW UNDER
DEVELOPMENT.
MEASURED
COEFFICIENTS
ASSIGNED TO
AVAILABLE.

OF EXPANSION WILL BE
EACH ARTIFACT WHEN

ELECTRICAL COMPONENTS DIVISION
SIDNEY NEW YORK 13838
TEST NO. 210495

PROCESS USED ΤΟ ESTABLISH THESE REFERENCE VALUES.

THE UNCERTAINTY IN ASSIGNED VALUE
CONTAINED IN THIS REPORT BECOMES A
SYSTEMATIC ERROR FOR THE LENGTH
MEASUREMENTS OF THE USER. IN THE
ABS ENCE OF OTHER SIGNIFICANT
SYSTEMATIC EFFECTS IN THE USER'S
MEASUREMENT PROCESS (A CONDITION
WHICH MUST BE DEMONSTRATED) THE
UNCERTAINTY OF THE VALUE ASSIGNED
BY THE USER IS AN APPROPRIATE
COMBINATION OF THE SYSTEMATIC
ERROR IN THE STANDARD AND THE
RANDOM COMPONENT ASSOCIATED WITH
HIS PROCESS. IF THE MEASUREMENT
PROCESSES ARE IN CONTROL AND
APPROPRIATE UNCERTAINTIES ARE
ASSIGNED, THE VALUES PRODUCED BY
DIFFERENT MEASUREMENT FACILITIES
WILL HAVE OVERLAPPING UNCERTAINTY
BANDS.
ONE CANNOT DISCUSS
DIFFERENCES IN VALUES FOR THE SAME
OBJECT OBTAINED BY DIFFERENT
FACILITIES WITH ANY DEGREE OF
SERIOUSNESS UNLESS EACH VALUE IS
ACCOMPANIED BY
REALISTIC
UNCERTAINTY STATEMENT.