CRANV1P1 Formation of Index Languages 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. - 83 - It has already been noted that the formidable task of adding to the 'one-place' schedules all other possibly useful hierarchies was not attempted. This was partly because much of the effort would have been wasted (if no questions were asked in- volving these alternative hierarchies), and partly because the A/Z index was likely to disclose the most important ones. It was also thought that the detailed analysis of reasons for failure (an integral part of the test programme) would disclose any examples of failure due to the absence of such alternative hierarchies. It should be remembered that the hierarchies actually established were those reflecting the most likely approaches to the material and that for many of the concepts alternative approaches (manifested in different citation orders) were quite obviously unnecessary; to take some at random for example, Mixture of cold gases could not conceivably enter into a search for kinds of mixtures, or kinds of cold things; the same applies to a number of other concepts involving the term Mixture. Similarly, in the case of a number of concepts involving the word 'modes', or 'models', it was unnecessary to contemplate the possibility of having hierarchies based on these (although hierarchies of particular kinds of model, e.g. , wind tunnel models, were of potential value, of course). The references already provided in the schedules and by a file of 'notes and decisions' assembled during the indexing were now supplemented by those in various thesauri and subject heading lists in the field of aeronautics and astronautics, since these are in principle the product of similar observation of connections between classes. Examples of such references are those from Heat transfer to Transport coefficients, Large Peclet numbe/" and Prandtl number; from Dissociated stream to Ionized boundary layer; from Kinetic theory to Diffusion and to Transport properties. It has already been noted that all such connections could, if necessary, be incorporated in a hierarchy of the kind being tested, although no distinction was drawn between generic and non-generic relations when utilizing these references. The combination of search programmes (10) and (11) represents, by and large, the performance of non-generic hierarchical relations largely, combined with a smaller element consisting of those:supplementary generic relations not shown direct- ly in the 'one-place' schedules. Although it has already been argued that both these hierarchical relations are generally quite secondary to the main display of generic relations, it must be regarded as a weakness of this joint presentation that separate programmes were not made for the two distinct situations. 3. Control by pre-established thesaurus A major objective in producing the concept hierarchies described in the last section was to afford a degree of precoordination sufficient to remove the artifici- alities accompanying the use of single words only in the 'one-place' index language and to provide where :suitable, that minimum of syntactical linkage necessary to the clear conveyance of unambiguous meaning in the index descriptions. It was thought that the resulting index language approximated more closely to the usual environment of index devices than did the first language. By this time, the search methods developed in the course of testing the first languages were producing the first detailed performance figures for the various devices and languages concerned. Although the operation of the large number of variables produced an extremely complicated picture in that many ways of aggregating these variables and their different values presented themselves, the general picture seemed to suggest clearly enough that the performances were not very encouraging.