An improved method is proposed for eliminating the commonly occurring spurious ripples which arise from interpolation fits to very accurate data and have an unknown functional form. This method is illustrated by fitting an interpolation curve to a commercial Ge resistance thermometer characteristic. For data with a rms scatter of ± 0.8 mK, an interpolation accuracy of about ± 0.2-0.4 mK is achieved. This accuracy is about four times higher than that obtained from the generally used high-order, polynominal, least-squares fitting method. It is also explicitly shown that the proposed method is superior to other interpolation methods.
Bibliographical noteFunding Information:
The authors express their appreciation to Prof Isaac Freund as well as to Profs Nathan Wiser and Marshall Luban for a critical reading of the manuscript. This work was sponsored in part by the Israel Commission for Basic Research under Grant No 12(D)P and by the United States - Israel Binational Science Foundation (BSF), Jerusalem, Israel under Grant No 342.