408 G. ADRIAN HORRIDGE 



of the efferent neurons and not depend on a specific central wiring diagram at all. 



(6) Simpler ganglia in the lower animals are homogeneous, rather than 

 divided into regions, and each neuron usually has branches which ramify 

 over the whole ganglion, intermingling with those of many other neurons. 



(7) As we go to more comphcated animals the neuropile is progressively 

 differentiated into more definite regions, and particular nerve cells have their 

 arborizations confined to one of these regions. However, their branches still 

 have the appearance of not being specified in detail and perhaps each region 

 now acts as I have supposed for a more primitive whole gangUon. 



(8) Within one animal, one gangUon may be highly regionalized with 

 exactly delineated spread of the axon arborizations and dendrites, as in the 

 optic ganglia of the dragonfly larva, another may be less regionalized, as in 

 the abdominal gangha of the same animal (Zawarzin, 1924) and another may 

 be at the simplest level, with little differentiation of neuron types or limitation 

 of the spread of neurons, as in the insect stomatogastric ganglia (Orlov, 1924). 



(9) The degree of localization and the directionality of the arborizations 

 range from a few microns, such as the endings of the retinula cell axons in 

 the optic ganglia of insects, to several milhmetres, or the whole extent of a 

 ganglion. Most arborizations lie somewhere about the middle of this range. 



(10) The pattern of arborization is rarely isotropic. The limitation of 

 growth in a particular direction, with a definite degree of dispersion, sets an 

 interesting problem in morphogenesis. 



(11) The degree of bushiness of the arborizations is typical of each neuron, 

 and frequently characteristic of large classes of neurons, e.g. in the corpora 

 pedunculata or the antennal lobes of insects many fine short branches are the 

 rule. A physiological correlate is not known. 



(12) The above features apply equally well to the dendrites of efferent 

 neurons and interneurons as they do to the axon arborizations of sensory 

 cells and interneurons, although one normally thinks only of the latter 

 in considering neurons with widespread and apparently non-specific effects. 



SUMMARY 

 A series of examples suggests that inhibitory and excitatory neurons in the 

 nervous systems of many invertebrates have widespread non-specific con- 

 nexions. The evidence is stronger where such systems are inhibitory because 

 then the afferent pathway must run to all spontaneously active neurons. In 

 general, non-specific systems are considered as the more primitive type, upon 

 which anatomical specificity of contact has evolved to various extents. Except 

 in a single nerve net, each neuron is addressed so that it transmits excitation 

 to some neurons rather than others which it meets. This addressing may be 

 anatomical, in which case synapses occur between recognizable neurons, or 

 physiological, in which case it may depend on differential sensitivities of recep- 

 tor neurons to transmitter substances released by other neurons. The effect of 



