CRANV1P1 Additional Tests chapter Cyril Cleverdon Jack Mills Michael Keen Cranfield An investigation supported by a grant to Aslib by the National Science Foundation. Use, reproduction, or publication, in whole or in part, is permitted for any purpose of the United States Government. - 112 - which have, for example, two references in common. However, proportionately, it could be argued that the latter represents a stronger match than the former. To test this, the number of references in each document of a matching pair were multiplied, the resultant figure was then divided by the square of the number of matches and the final figure was considered as the level of coupling strength. For example, document 1067 (see Fig. 7.4} had six references, and document 1163 had eight references, giving a multipland of 48. These had a match of 2, so dividing by 22 gives a final weighted figure of 12. Document 1715, however, had nine references which, combined with document 1067, gives a multipland of 54. In this case, since there is a match of four, this figure has to be divided by 42, giving a final weighted figure of 3. When the matches for document 1067 had all been worked out, the weighting becomes as in Fig. 7.6. In many cases, the result of this exercise showed significant changes in coupling strength, and therefore the collection was re-tested in the manner described earlier, only this time the scoring was based on these new coupling levels.