396 G. ADRIAN HORRIDGE 



On the other hand we are famihar with the attitude that the nervous system 

 is a biological structure which by virtue of patterns of growth is presumably 

 to some extent ordered, and only as an unfavoured alternative are theories 

 developed to show how a random set of connexions can provide, in a formal 

 way, and with economy in number of assumptions, an adequate model of 

 some aspects of the structure. This is understandable but abandoned in the 

 present article for two reasons. Firstly, it is the least complex explanation of 

 both the structure and its behaviour which is sought, and the degree of com- 

 plexity is judged by the number of specifications required to formulate the 

 model which takes account of both aspects. Secondly, it appears to the 

 writer that the increase in complexity of the invertebrate nervous system, 

 especially at the lower end of a series from coelenterates up to arthropods, 

 can be profitably considered as a progressive but incomplete development 

 away from a random structure towards a physiologically or anatomically 

 more ordered structure. If this is so, then the process has necessarily been 

 accompanied by a differentiation of the neurons into distinct classes and in 

 many instances, especially on the efferent side, into unique individual neurons. 

 However, as will be shown, specific anatomical contacts between potentially 

 recognizable distinct neurons are only one method of achieving this differ- 

 entiation of specific interrelationships. 



Together with definitions of some of the terms which I shall use, it is 

 helpful to consider the anatomical form of a neuron by analogy with the path 

 taken by a letter during the delivery of mail. Letters which are addressed to a 

 single recipient may take a circuitous path to their destination and similarly 

 addressed letters from the same source may go by different paths. Similarly, 

 a recognizable neuron, when found again in a different segment or a different 

 animal, need not have exactly the same anatomical form but may neverthe- 

 less be in part "anatomically addressed" in so far as it makes particular 

 contacts with other recognizable neurons. The recognizable and therefore 

 regularly discernible structural relationships of a neuron with the others which 

 it excites are defined as anatomically addressed. In the anatomically addressed 

 maihng system the individual letters need not necessarily have any content, 

 and information can be transmitted by the frequency of arrival of empty 

 envelopes. 



In a different but not necessarily mutually exclusive system, letters do not 

 bear individual addresses but are distributed with various degrees of approxi- 

 mation as to the area of their destination, as when leaflets are scattered from 

 the air, but with definable shapes and extents of the area covered. This 

 situation is similar to that of a branching neuron which makes contact with 

 any other neuron it happens to meet in the area of its terminations. As in 

 the case of the models of the coelenterate nerve net, the leaflet in the mail 

 analogy can be a blank sheet which by its arrival merely triggers a response, 

 and, as in the anemone nerve net, in particular, the frequency of arrival of 



