CRANV1P1 Indexing Procedures 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. -57 - being developed. This was the problem of knowing what terms were qualified by what adjectives. The example of 'High aspect ratio wings at low angles of attack' has already been cited. Without some interlocking device (links or roles) this docu- ment description would respond to a question on 'Low aspect ratio wings', or 'High angles of attack'. Similarly, an index description 'Transient aerodynamic heating of external loads' might respond to a question on 'Transient loads' or 'External heating', and '[OCRerr]Compressible flow over an adiabatic wall' might respond to a question on 'Adiabatic flow'. However, the situation is typically one which is solved by links rather- than roles, since the type of relation is not in dispute, only which of the two (or more) terms are to be linked. If such modifiers were to be regarded as roles the only solution which would remove__all ambiguity would be to give them the same role as the term they modified, and then it would work only on the assumption that links were used as well. This was therefore tried (using such roles as Specifier of agent, Specifier of structure), but only at the cost of duplicating the linking function already performed by partitioning and interfixing. One of the few situations, in the ]iterature indexed in which a concept (rarely a single term) might feature in a role which appears to differ significantly from its usual one is that in which a body of data is used as an agent in the investigation of some other phenomenon (to which it therefore takes second place in the document concerned). For example, in B1204,Two-dimensional airfoil section data are used as a means of studying high subsonic speed characteristics of swept wings. But closer examination throws doubt on the necessity for using a role even here; had the section data been referred [o in some other way (as a parameter influencing aero- dynamic behaviour, say) its essential relation to the primary subject (swept wings) would not appear to have been altered. It was stated earlier that in transferring concepts to themes (the second step in indexing) the exact relationship was sometimes indicated for clarity. The relative infrequency with which the need to do this arose (and Document 1590 was not typical here) was itself a warning that the field of high speed aerodynamics was not likely to be very fruitful in the evaluation of roles as a device. This impression was rein- forced by our preliminary analyses. Since, then, any measure of roles was likely to reflect an uncongenial ar, d unresponsive test environment, the very considerable effort involved in developing a set of appropriate roles, applying them and measuring the impact, seemed of doubtful worth. For this reason the preparations for the evaluation of roles as a device were temporarily abandoned.