[In the following review, Young observes the significance of Ramóny Cajal's work to modern neuron theory.]

Those who probe the nervous system with electrodes probably seldom stop to consider the history of knowledge of the cells they are impaling. Yet it would help them to think about the problems that have arisen in the search for units of nervous activity. Since the days of Santiago Ramón y Cajal, most neuroscientists have depended on a rather simple picture of the neuron, with dendrites, cell body and axon as the essential unit. This has also been the model mostly used in artificial intelligence. Shepherd's book [Foundations of the Neuron Doctrine] provides a survey of the history of the neuronal hypothesis. In his last chapter, he raises the question of whether we should now look for units both larger and smaller than the neuron.

The controversy at the end of the last century turned on the question of whether neurofibrils proceed from one cell to the next. It was conducted in fairly ferocious language. Cajal writes of his reticularist opponents, such as Golgi and A. Bethe, as "fanatics with haughty minds, inclined towards mysticism". Finally, in 1917, he is happy to write that "the unhorsed physiologist of Strasbourg [Bethe] decided to abandon the field. Victis honos!"

Cajal was, of course, correct in claiming that "connection is by contact", but his opponents were skilful light microscopists and not so far wrong as he supposed. Now that electron microscopy has shown the correct relationships at synapses, we can see that their interpretations were in a sense correct. The finest branches of a nerve fibre may indeed appear to enter the end organ, for instance in the groove at the surface of a muscle fibre. There is no evidence that Cajal realized that it is the completeness of the two membranes that is important. If the finer branches run in a trough, the most honest light-microscope interpretation may be that there is continuity. Cajal's opinion was right, but his figures are almost all drawings.

The advocates of the neuron theory were themselves quite "haughty" and hasty in their rejection of all possibilities of "continuity". We know now that gap junctions may allow passage of ions and small molecules between neurons. Furthermore, there may be complete fusion of nerve cells if they always function together. For instance, the two giant cells of the squid initiate contraction of the muscle sac—and they are completely joined by a bridge: for jetting, both sides of the mantle must contract together. But where impulses are initiated there are synapses. This is a system of "Fused neurons and synaptic contacts", as the paper in which it was described was called in 1939. The fusion is the exception that proves the rule. Nerve fibres can fuse, but where decisions are to be made they are separated by synapses. I remember explaining all this to Sherrington (in about 1938). He looked up at me quizzically and said, "I hope that you are right Young, but I find it hard to believe." It is ironic that the squid's giant fibre synapse, more thoroughly investigated than any other, involves a syncytial postsynaptic fibre. I hope that Cajal would have enjoyed the joke (but I'm not sure that he would).

This history of old doubts and quarrels shows how hard it is to arrive at secure knowledge. As more has been discovered it becomes clear that the classical neuron doctrine needs to be extended. Almost from the start there were doubts as to what the term should include. The word 'neuron', originally suggested by Waldeyer in 1891, comes from the Greek, meaning, literally, tendon or sinew, and was applied through confusion to nerve trunks. Some authors therefore wished to keep the term neuron for the axon, whereas others (paradoxically) tried to use it only for the nerve cell body. Kölliker and others emphasized that the word should be spelled 'neurone'. This usage is still insisted on by some British physiologists and by Cambridge University Press. Shepherd nowhere mentions the history of this spelling. Many people must be puzzled to know which form to use and the book could have given authorative guidance. Surely British physiologists and Cambridge University Press should now abandon this pretentious and unhelpful practice and follow the rest of the world.

More serious are the problems raised by the discovery that dendrites may have synaptic outputs and that axons can have inputs from other axons. Moreover, dendrites do not always have graded synaptic responses but may carry voltagegated propagated action potentials, whereas, conversely, some axons do not carry these at all. Shepherd summarizes the effects of such processes: "there is not a fixed correlation of structure and function within the different parts of the neuron; axons and dendrites provide flexible substrates in which a variety of membrane channels and local organelles . . . can support different types of physiological properties and function operations. . . . So although the neuron remains a basic anatomical, physiological, genetic and metabolic unit .. . it contains several levels of local subunits, and is itself a part of larger multineuronal units." Such a neuron has several potentially modifiable parts. It will provide a truer picture for neuroscientists and theoreticians who are trying to model parallel computing systems (although there will be difficulty in constructing them).

It is useful to have, at the same time as this review of the neuron doctrine, a new issue of Cajal's own book on degeneration and regeneration of the nervous system. This was first published in Spanish in 1913-14, the cost of publication being covered by expatriate Spanish physicians in Argentina in honour of Cajal's Nobel prize. It was translated into English in 1928, but without several important sections that are included in the present edition. Complete with Cajal's excellent pictures of his preparations, the new edition makes a wonderfully full account of regeneration. Several of his most important ideas are developed here. Particularly relevant to modern work are the concepts of neurotropism and his studies of regeneration in the central nervous system. It is good to have this book, but it tempts one to complain that Cajal's greatest work, Histology of the Nervous System of Man and the Vertebrates, published in 1899 in Spanish and in 1909 in French, is still not available in English. Illustrated again by Cajal's beautiful figures, that book provides detail of every part of the brain and peripheral nervous system, and should be accessible to every neuroscientist.