CRANV2 Aslib Cranfield Research Project: Factors Determining the Performance of Indexing Systems: Volume 2 Test Environment chapter Cyril Cleverdon 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. CHAPTER 2 Test Environment Communication is the means which enables society to adjust itself to alterations of technology and education and other social changes. The scientific method can offer no grand vision, no global strategy, no panacea. It will never be possible to demonstrate that anything is absolutely right or even completely scientifically true. L.T. Wilkins: Social Deviance, page 28. In the first volume were considered the general plan of the test design, the variables that were to be investigated and the methods to be used, In the course of the project, changes were made regarding certain details, and this chapter presents the environment in which the testing was actually done. While an information' retrieval system may be defined in its scope as 'all stages from the receipt of a document within a system, to the making of that document (or a representation of it} available to an enquirer', not all these stages have been included in the investigations in the present project. The central concern was the effect of index language devices on the operational performance, but in addition a number of other variables or factors have been included for various reasons. In order to clarify later discussions, a breakdown of an indexing system into four main groups is suggested, namely environmental factors, software factors, operational factors and hardware factors (see Fig. 2.1}. The environmental factors relate to the environment or conditions in which a given system has to operate. Four general factors are given, and, in the case of an operational system, they are all determined to a great extent by the needs of the user group which the system exists to serve. The subect field, the questions asked and the relevance needs directly depend on the users, while the collection size will be determined by the management largely in relation to user needs. However, for an