The design envisaged was perceived as a compromise between three major 'orientations': production of a practical classified directory; facilitation of experiments on classifications to develop improved versions; and an emphasis on incorporating richer patterns of relationships between activities to facilitate understanding of functional integration. These are detailed separately below.

1. Practical orientation

In the light of the above survey, the factors affecting the design of a practical system may be summarized as follows:

• a) it should respond to the progressive increase in number of organizations with multi-subject concerns;
  b) it should meet the need for a relatively simple classification scheme;
• c) it should facilitate incorporation of changes in organizational activities with the emergence of new issues (environment, energy, etc);
• d) it should avoid the production delays associated with conven tional methods of classification, particularly with increasing numbers of organizations and with the change in their concerns;
• e) in order to facilitate solutions to the above problems, it should use an approach which could be assisted by computer techniques as much as possible;
• f) finally, and perhaps of greatest importance, it should result in the production of a practical directory which avoids confronting the average user with levels of significance or complication not required, even though these features may be present for those who wish to benefit from them.

2. Experimental orientation

In contrast to most current classification systems, the design should facilitate classification experiments in the light of the following factors:

• a) it was not intended to produce immediately a 'definitive' clas sification scheme for international organization activities;
• b) it was expected that different approaches will be explored from edition to edition, possibly with several approaches in one edition;
• c) the position of classes or sub-classes in any one matrix pattern might be adjusted between editions in the light of the results to which it gave rise when tested on the range of international organization activities;
• d) it was expected that refinements to the computer programmes used would lead to more valuable versions of the scheme;
• e) the flexibility necessary for such an experimental approach should be achieved by computer-assisted methods of reclassifying the complete range of organizations whenever a new version of the scheme is required;
• f) as an experimental system, risks would necessarily be taken which might give rise to errors, but every effort would be made to minimize their significance for users interested only in the practical value of a given classification scheme.

3. Pattern building orientation

It is hoped that experiments in classifying international organization activities will be carried out to highlight significant patterns of relationships between them in the light of the following factors:

• a) an emphasis less on possible bilateral relationships between any two subject areas (eg medicine and sport) as on portraying the complete range of classes in some functionally meaningful pattern of relevance to organization activities;
• b) the intention to explore ways of ordering the classes within as many simultaneously interweaving patterns as proves feasible;
• c) in developing such patterns a major constraint is that of maintain ing and improving the comprehensibility of any such scheme.