The meaning of the average of | F | 2 for large values of the interplanar spacing

constructed according to the general designs suggested by Lipson & Beevers (1936), Beevers & Lipson (1952) or Patterson & Tunell (1942). Table 2 compares the numerical values of 18 structure factors as computed with strips accurate to 0.01 and to 0.001. The parameters are again those given in Table 1, and the formulas for F(Okl) and F(hkO) with k+l odd are given by equations (3) and (4). Columns B and C compare directly the approximate numerical values derived with the small set of strips with the exact values computed using the large set for parameters accurate to three places. In view of the anticipated discrepancies at larger indices, the quality of the agreement is rather surprising. Even for the planes (630), (830) and (047) the errors need not cause concern save in the final stages of the parameterrefining process. The only logical explanation for this unexpectedly good agreement at the higher indices is that the errors in the trigonometric products for the various atoms, being random in sign and magnitude, tend to cancel each other on the average. For this reason the small set of strips may have a special value in solving complex structures involving many parameters. The lack of significant differences between the values in columns A and B demonstrates that in the earlier stages of parameter refinement one may assume the constancy of fc/fo and frr/fo with impunity.