398 G. ADRIAN HORRIDGE 



neurons run distinguishes from other classes of excitation. The present issue 

 is to suggest that besides the path of progressive differentiation of morpho- 

 genetically determined specific connexions, the evolution of labelled lines from 

 the primitive non-addressed system can include the progressive differentiation 

 of the substances secreted by the neurons and their sensitivity to secretions 

 of their neighbours. Labelled lines are neurons which have differentiated 

 along distinctive paths of development, either in structural or physiological 

 features, or in both. The problem of analysis is complicated because these 

 two modes of growth of complexity are progressive and not mutually ex- 

 clusive and it is possible that physiologically addressed diffusely spreading 

 neurons act among and together with other neurons having various degrees 

 of specificity of anatomically addressed connexions. Also, a physiologically 

 addressed neuron must to some extent be addressed anatomically to a parti- 

 cular region and an anatomically addressed one has specific modes of chemo- 

 sensitivity. A physiologically addressed system must have a neuropile-like 

 widely ramifying structure but an anatomically addressed system need not be 

 arranged this way. 



Neuropile is a word which I shall use for the intermingled processes of 

 dendrites and axon arborizations which branch profusely and ramify among 

 each other in the central nervous systems and outlying ganglia of all animals. 

 It is characteristic of nerve cells that where they meet and apparently generate 

 all interesting aspects of behaviour, they produce this profuse branching and 

 relatively great surface area with muhilateral relationships with other neurons 

 in all directions. In the vertebrates there are regularly, in addition, endings 

 on cell bodies, but all invertebrates have nervous systems where all the 

 interesting activity seems to be in the neuropile, as defined in the above way, 

 and a similar neuropile occurs regionally in vertebrates, as, for example, in 

 the optic tectum of the lower forms and the granular layers of the cortex of 

 mammals. 



A central problem of invertebrate neurophysiology is the analysis of 

 systems of this type, leading to the understanding of why there should be this 

 dense intermingling of many profusely branching fibres. However, present 

 techniques do not provide the crucial information. The physiological data 

 now available from several convenient preparations of invertebrates, parti- 

 cularly gangha of arthropods and molluscs, can be represented diagram- 

 matically by sets of simple connexions between stylized neurons where each 

 acts on the next down the fine. But neuropile in fact does not look Uke this, 

 and from anatomical studies of neuropile no one has ever obtained anything 

 which looks like a wiring diagram. 



The present ideas leading to a new approach spring from a series of obser- 

 vations on animals which have many redupHcated pacemakers, all of which 

 can be inhibited at the same time. The simplest such systems occur among 

 those coelenterates which have more than one nerve net. In some coelenterates, 



