LECTURE IICO-ORDINATION IN THE SIMPLE REFLEX (continued)

Summation. Summation of subliminal stimuli so that by repetition they become effective is practically unknown in nervetrunk conduction. But it is a marked feature of reflex-arc conduction (Setschenow,³¹ Stirling⁵⁷). Nor is it attributable to the muscles whose contraction may serve as index of the reflex-response, since summation of this extent is not known for vertebrate skeletal muscle, though found by Richet⁹⁸ in the claw-muscle of the crayfish.

We find striking instances of the summation of subliminal stimuli given by the scratch-reflex. The difficulty in exciting a reflex by a single-induction shock is well known. A scratch-reflex cannot in my experience be elicited by a single-induction shock, or even by two shocks, unless as physiological stimuli they are very intense and delivered less than 600 σ apart. Although the strongest single-induction shock is therefore by itself a subminimal stimulus for this reflex, the summating power of this reflex mechanism is great. Very feeble shocks, each succeeding the other within a certain time — summation time — sum as stimuli and provoke a reflex. Thus long series of subminimal stimuli ultimately provoke the reflex. I have records where the reflex appeared only after delivery of the fortieth successive double shock, the shocks having followed each other at a frequency of 11.3 per second, and where the reflex appeared only after delivery of the forty-fourth successive make shock, the shocks having followed at 18 per second. A momentary stimulus, e. g, a break shock of fair physiological strength applied by a stigmatic pole (needle point) to a skin-spot in the receptive field of this reflex, produces in the nervous arc a change which though, as just said, unable of itself alone to produce the reflex movement, shows its facilitating influence (bahnung) on a subsequent stimulus applied even 1400 σ later. The duration of the excitatory change induced by a momentary stimulus is therefore in this mammalian arc (scratch-reflex) almost as long as that noted in the frog by Stirling, namely, 1500 σ.

With serial stimuli of the same frequency of repetition the latent time of the scratch-reflex is shorter the more intense the individual stimuli. Stirling⁵⁷ conclusively traced length of latency to dependence on spinal summation of successive excitations. In accord with this in the “scratch-reflex,” when the serial stimuli follow slowly, the reflex caeteris paribus is prolonged. A single brief mechanical stimulation of the skin (rub, prick, or pull upon a hair) usually succeeds in exciting a scratch-reflex, though the reflex thus evoked is short; but there is nothing to show that these stimuli, though brief, are really simple and not essentially multiple. A striking dissimilarity, therefore, between reflex-arc conduction and nerve-trunk conduction is that in reflex-arc conduction considerable resistance is offered to the passage of a single nerve-impulse, but the resistance is easily forced by a succession of impulses; in other words, subliminal stimuli are summed.

It follows almost as a corollary from this that the threshold excitability of a reflex mechanism appears much more variable than that of a nerve-trunk, if the threshold excitability be measured in terms of the intensity of the liminal stimulus. The value will be more variable in the case of the reflex mechanism, because there the duration of the stimulus is a factor in its efficiency far more than in the case of the nerve-trunk. In the scratch-reflex a single stimulus which is far below threshold intensity is found, on its fortieth repetition and nearly four seconds after its first appplication, to become effective and provoke the reflex.

Irreversibility of direction of conduction. Another remarkable difference between reflex-arc conduction and nerve-trunk conduction is the irreversibility of direction of the former and the reversibility of the latter. Double conduction, as it has been termed, is well-established for nerve-trunks both afferent and efferent. It was shown by du Bois Reymond for the spinal nerve roots, for peripheral nerves by Kuhne’s gracilis experiment, for the great single electric fibre of Malapterurus by Babuchin,⁷³ for sympathetic nerve-cords by Langley and Anderson,¹⁵⁰ and by myself¹⁷¹ for certain fibres of the white tracts of the spinal cord. The nerve-fibres in all these cases, when excited anywhere in their course, conduct nerve-impulses in all directions from the point stimulated; that is, in their case both up and down, the only two directions open to them. Their substance may therefore be regarded as conductive in all directions along their extension.

From the Bell-Magendie law of the spinal nerve-roots we know that reflex-arcs conduct only in one direction. The stimulation of the central end of a motor-nerve remains without obvious effect. Bell¹⁰ and Magendie¹¹ and their followers established that excitation of the spinal end of the severed motor root evokes no sign of reflex action or sensation. Evidently the central nexus between afferent channel and efferent is of a kind that, though it allows conduction from afferent to efferent, does not allow it from efferent to afferent. The path is patent in one direction only. This is the special case which forms the first foundation of the law that conduction in the neural system proceeds in one direction only, the “law of forward direction (W. James, 1880).⁸² When the property of double conduction in nerve-fibres had been ascertained, the Bell-Magendie law of the spinal roots became more instructive. Gad¹⁰² (1884) argued that the dendrites of the motor root-cell are capable of conduction in one direction only, namely, toward and not away from the axone. It may, however, be that the irreciprocity of the conduction is referable to the synapse. The explanation of the valved condition of the reflex circuit may lie in a synaptic membrane more permeable in one direction than in the other. In other words, through intraneuronic conduction is reversible in direction, interneuronic may be irreversible.

Cell-chains of polarized conduction form the basis of the great majority of all the nervous reactions of the cerebrospinal system of higher animals. It appears, however, that not all pluricellular nervous circuits exhibit irreversible direction of conduction. The nerve-net of Medusa is a pluricellular conductor which exhibits reversibility of direction of conduction. In Medusa locomotion is effected by contraction of a sheet of muscle in the swimming-bell. When the swimming-bell, which resembles an inverted cup, contracts, its capacity is lessened, and some of the water embraced by it is expelled through the open end, the animal itself being propelled in the reverse direction by recoil. The mechanism is like that of the heart, but the heart propels its contents, the swimming-bell propels itself against its contents. The contractions of both recur rythmi-cally, though Medusa, unlike the heart, has periods of prolonged diastolic inactivity. At such a period an appropriate stimulus restarts the swimming-bell. The contractive beat begins from the point stimulated and spreads thence over the whole muscular sheet.⁷² It spreads rapidly enough for the contraction not to have culminated at the initial point before it has set in at the most remote part. The beat is thus not only everywhere in progress at the same time, but is practically in the same phase of progress everywhere, and similarly synchronously passes off.⁷²

The arrangement of the nervous system of Medusa, e. g., Rhizostoma, is, according to Bethe, of the following kind (Fig. 11). The nerve-cell has on one hand thread-like arms that extend to the surface of the subumbrella, and on the other hand others which stretch down to the sheet of contractile cells on the under side of the bell. Each nerve-cell has also long side threads which join similar side threads from other nerve-cells. By virtue of these lateral connections the nerve-cells form a network of conductors spreading horizontally through the bell in a layer of tissue between a receptive sheet and a contractile sheet. From this nerve-net, throughout its extent, there pass nerve-threads to the adjacent muscle; it also receives at many points of its extent nerve-threads from specially receptive areas of surface.

The circularly arranged sheet of muscle does not form a continuous field toward the centre of the disc; there are wide radial gaps in it. Across these gaps the “conduction” passes: the microscope reveals no muscular tissue in these gaps, but the nerve-net can be seen to spread across them.²⁶³ The presence of the nerve-net explains the conduction across them. It is therefore argued by Bethe that the spread of the contraction over the muscular sheet in Rhizostoma does not imply conduction of the contraction from one muscle-cell to another, but is the result of the spread of nervous action over the nerve-net work. In its progress along the nerve-net, the nervous discharge, as it reaches each part of the nerve-net, spreads down the nerve-threads, descending thence to the underlying muscle-sheet. So long as the nerve-cell network is intact, wherever the point stimulated, the ensuing contraction is of the whole bell, that is, the nerve-impulses started at one point of the receptive surface, on entering the nerve network, spread over it in all directions. When the bell-shaped disc is spirally cut into a long band, to whichever end of the band the stimulus be applied, the conduction spreads from that end to the other and over the whole strip (Romanes).⁷² The nerve-net therefore conducts nerve-impulses in both directions along its length. Therefore it is not a polarized conductor, conductive in one direction only. In the chains of nerve-cells of higher animals, such as Arthropods and Vertebrates, although the conduction is reversible in each nerve-cell, — at least along that piece of it which forms a nerve-fibre, — the pluricellular chain in toto constitutes a polarized conductor, conductive in one direction only. In such cell-chains the individual nerve-cells are characterized morphologically by possessing two kinds of cell-branches, which differ one from another in microscopic form, the one kind dendrites, the other axones. The difference in appearance between dendrites and axones is marked enough for recognition by microscopical inspection. Since in many well-known instances the dentrites conduct impulses away from their free ends, while the axone conducts towards its free end, it is possible on mere microscopic inspection of nerve-cells of this type to infer by analogy the normal direction of the conduction through the nerve-cell. But in the nerve-cells forming the nerve-network of Medusa there seems no such distinct differentiation of their branches into two types. Their cell-processes are not distinguishable into dendrites and axones.

Moreover, microscopic examination of the nerve-net of Medusa reveals another difference between it and the nerve-cell chains of higher animals. In these latter the neuro-fibrils of one nerve-cell are not found unbrokenly continuous with those of the next cell along the nerve-chain. Although the union may be close, there is not homogeneous continuity. The one nerve-cell joins another by synapsis. But in the nerve-net of Medusa the neuro-fibrils pass, according to Bethe, uninterruptedly across from one cell to another. Even if we admit the neuro-fibrils to be in a measure artifacts, the appearance of their continuity from one cell to another in one type, and of their discontinuity from one cell to another in the other type remains significant of a difference between the conduction-process from cell to cell in the two types. The nerve-net of Medusa appears an unbroken retiform continuum from end to end. Each nerve-cell in it joins its neighbours much as at a node in the myelinate nerve-fibre the axis-cylinder of each segment joins the next. Reversibility of conduction may be related to this apparent continuity of structure, and irreversibility to want of it. This points to the latter’s being referable to the synapse; if the synaptic membrane (Lect. I. p. 18) be permeable only in one direction to certain ions, that may explain the irreversibility of conduction. The polarised conduction of nerve-arcs would be related to the one-sided permeability of the intestinal wall, e.g. to NaCl (O. Cohnheim).

Rhythm of response. One of the differences between nerve-trunk conduction and reflex-arc conduction is the less close correspondence in the latter between rhythm of stimulus and rhythm of end-effect. The number of separable excitatory states (impulses) engendered in a nerve-trunk by serially repeated stimuli corresponds closely with the stimuli in number and rhythm. Whether the stimuli follow each other once per second or five hundred times per second, the nervous responses follow the rhythm of stimulation. Using contraction of skeletal muscle as index of the response the correspondence at rhythms above thirty per second becomes difficult to trace, because the mechanical effects tend at rates beyond that to fuse indistin-guishably. The electrical responses of the muscle can with ease be observed isolatedly up to faster rates: their rhythm is found to agree with that of stimulation; thus, at eighty per second their responses are eighty per second. If the muscle note be accepted as an indication of the response of the muscle, its pitch follows pari passu the rate of stimulation of the nerve through a still greater range.

The case is quite different with reflex-arcs. Schäfer²⁰⁶ noted undulations of a frequency of ten to twelve per second on myo-grams of spinal reflexes evoked by excitation of the afferent nerve by faradic currents of frequency much above ten to twelve per second. In such a case we may assume the absence in the afferent nerve itself of any refractory period long enough to give a ten per second rhythm to the response. The refractory period in nerve-trunk conduction seems to last not longer than 1 σ. The rhythm of discharge from the motor-cell, as far as the undulations noted indicate rhythmic response, are totally different in rhythm from that of the action induced in the afferent cell by the stimulation applied. In the reflex centre the rhythm has been transmuted from one rate to another. Schäfer refers this change to the synapse.

Again, as noted above, undulations at rates varying between 7·5 and 12 per second are seen in the flexion-reflex both in its after-discharge and during the excitation and quite independently of the rate of delivery of the induction shocks used as stimuli (Figs. 6 and 12), and even when the stimulus is a constant current. Again, reflexes of weak intensity, both in the case of the flexion-reflex and of the crossed extension-reflex, exhibit at their commencement a stepped form of myogram: the steps usually succeed each other at about eight per second, and their rate is independent of the rate of delivery of the electric stimuli (Fig. 12) exciting the reflex.

In such cases, therefore, the rhythm of the end-effect indi

cates that in transmission along the reflex-arc the impulses generated at the receptive end of the arc are not actually passed on from one cell element to another in the arc, but that new impulses with a different period are generated in the course of the reflex-conduction. This is confirmatory of a neurone-threshold as a feature in central conduction.

Refractory phase. A conductor which replies intermittently to a stimulus exhibits a refractory phase. It seems that even in nerve-trunk conduction a refractory phase must occur; otherwise, the conductor being capable of reversible direction of conduction, a backward propagation of the excited state as well as a forward would ensue from every point of the conductor reached by the nervous impulse. The excited state would then, when once excited, maintain itself in a tetanic manner along the whole length of the conductor. The propagation would thus lose undulatory character such as we know it has; it would merely have an initial wave-front. But this refractory phase in nerve-fibres seems of very brief duration, not longer than 1 σ.

The reflex-discharge from nerve-cells seems to be rhythmic even under continuous stimulation; but the cases in which this has been examined are sparse. The existence of rhythm in nerve-cell discharge is presumptive evidence of a refractory phase in their reaction. Refractory phase was first called attention to by Kronecker and Stirling⁵⁵ in 1874, in the heart, and recognized by them as a fact of central importance for cardiac rhythm. In 1876 Marey⁶⁷ met the same phenomenon and gave it the name by which it is now known. A year later Romanes’ fundamental work on Medusa demonstrated the existence of the same phenomenon there.⁷² The inconspicuous duration of the phase in nerve-trunk conduction and the progress of the view that regards the heart beat as of myogenic origin have contributed to delay recognition of refractory phase as a character of reflex-arc reactions. But in 1899 Zwaardemaker and Lans¹⁹⁸ showed the supervention of a refractory phase in reflex eyelid-closure. By refractory period was originally meant by Marey the time during which the heart was inexcitable to a stimulus however intense. But to-day by refractory phase is understood a state during which, apart from fatigue, the mechanism shows less than its full excitability. The cardiac refractory phase is absolute for a short time after commencement of systole, but excitability then returns gradually. In the eyelid-reflex for nearly a full second after initiation of a reflex, the chance that a second stimulus then delivered will, though otherwise appropriate, excite the reflex, is fifty per cent less than it is one second later. The refractory phase, therefore, is marked though not absolute: it operates longer for a visual stimulus than for a tactual or thermal.

Of the spinal reflexes of the dog’s hind limb, one, namely, the scratch-reflex, shows marked refractory phase.²⁸⁷ When working originally with this reflex, I noticed²⁵¹ that the rhythm of the scratching movement is in rate independent of the rhythm of the stimulus evoking it.

In the dog, when the spinal cord has been transected in the neck, the scratch or scalptor reflex becomes in a few months prominent. A stimulus applied at any point within a large saddle-shaped field of skin (Fig. 13) excites a scratching movement of the hind leg. The movement is rhythmic alternate flexion and extension at hip, knee, and ankle. Each flexion recurs at a frequency of about four times per second. The stimuli provocative²⁵² of it are mechanical, such as tickling the skin or pulling lightly on a hair. The receptive nerve-endings which generate the reflex lie in the surface layer of the skin, about the roots of the hairs. A convenient way of exciting the reflex is by feeble faradization, such as applied to one’s own tongue is felt as a tickling sensation. For exciting the reflex electrically, I place a broad diffuse electrode on some indifferent part of the surface outside the receptive field of skin, and apply a stigmatic electrode to some point in the saddle-shaped area of dorsal skin. This electrode may consist of a minute needle, or a gilt entomological pin; it is inserted in the skin so lightly that its point just lies among the hair bulbs. If it be pushed further, other types of reflex may be produced and the scratch-reflex be inhibited. Prominent among the muscles active in this reflex are the dorso-flexors of the ankle, the flexors of the knee, and the flexors of the hip. If the rhythm of the last is recorded graphically, tracings,³⁰⁰ as in Figure 14, are obtained. It is then demonstrable that the series of brief contractions succeed one another at a rate the frequency of which is independent of that of the stimulation. Thus, the rhythmic reflex is elicitable by the application of a heat-beam or a constant current. The make and break of the current are especially able to excite it (Fig. 15), but it is for a time maintained even by the continued passage of the current, though it lapses in intensity until the current is broken, when it appears temporarily with renewed vigor. At make and break of the voltaic current the reflex-response is a short series of rhythmic flexions.

The reflex is still more easily evoked by unipolar application of high frequency currents; it can be evoked with great vigour and long duration by this mode of stimulation. Double-induction shocks applied at frequencies from once per second to five hundred and twelve times per second and at various intermediate rates all evoke it easily; so likewise do single break or make shocks applied at rates varying in my experiments between 1·33 times per second and forty times per second.

Under all these various methods of excitation (heat-beam, constant current, double and single induced currents, high frequency currents, and mechanical stimuli) the rhythm of the flexor response remains — so long as the internal conditions of the reflex remain unaltered — almost the same.²⁸⁷ It remains so also when, instead of a regular succession, a grouped succession of stimuli is used for excitation, e. g., stimuli grouped in twos and threes. It is obvious that this reflex exhibits a refractory phase. Take the instance where the stimulus applied consists of double-induction shocks, succeeding each other at a frequency of one hundred per second. The reflex-arc in response produces flexion at hip about four times per second. So few as three successive double-induction shocks will suffice to excite the reflex, but since in the instance taken about twenty such shocks correspond in time with each flexor beat, and each such beat is divisible into about equal periods of contraction and relaxation, let us make the assumption — a liberal one — that ten out of the twenty available shocks are serving as stimuli. The arc, while the following ten are being applied, fails in spite of them to excite the flexor muscles. To those ten it does not respond at all.

Nor can this refractory state be overcome by simply increasing the intensity of the stimuli. The reflex remains as perfectly rhythmic and clonic under the strongest stimuli as under weaker (Fig. 16). The frequency of the beats is under strong stimulation often somewhat higher than under quite weak, especially at outset of the reflex, but the difference is small, e. g., 5·8 beats per second, instead of 4·5 beats per second. No mode or intensity of stimulation to which I have had recourse converts rhythmic clonic beat into a maintained steady contraction. In this respect, as in certain others, the scalptor reflex-arc closely resembles in its behaviour the mechanism of the Medusa bell and heart-wall. It resembles these more closely than it resembles the mammalian respiratory reflex-arc, which can under several circumstances be made to produce from the diaphragm an enduring tetanic contraction.

I refer to the contraction of the flexor muscles in the scalptor-reflex as a “beat,” not implying that it is a single twitch, but rather because the term denotes a short-lasting phase of action in a rhythmic series, and because of its close analogy to the beat of Medusa and of the heart.

What is the neuronic construction of the arc in the scalptor-reflex? The reflex is in a sense unilateral; stimulation of the left shoulder evokes scratching by the left leg, not the right. Search in the spinal cord²⁵¹ for the paths of the reflex demonstrates that a lesion breaking through one lateral half of the cord anywhere between shoulder and leg abolishes the ability of the skin of that shoulder to excite the scratch-reflex, but leaves intact the reflex of the opposite shoulder. In the lateral half of the spinal cord which the reflex-path descends, severance of the dorsal column does not obviously interfere with the reflex; nor does the severance of the ventral and the dorsal columns together of that side; no more does severance of the gray matter in addition. But severance of the lateral part of the lateral column itself permanently abolishes the conduction of the reflex; and it does so even if all the other parts of the transverse extent of the cord remain intact. The paths of the reflex, therefore, descend the lateral part of the lateral column. These details help towards construction of the reflex-arc involved. For in the lateral part of the lateral column, as shown by the method of successive degeneration, lie long proprio-spinal fibres which directly connect the gray matter of the spinal segments of the shoulder with the spinal segments containing the motor neurones for the flexor muscles of the hip, and knee, and ankle. The course of the descending proprio-spinal fibres can be traced and their number counted. The method of “successive degeneration”²⁵¹ enables one to unravel them from descending fibres of other sources, such as cerebral, mesencephalic, or bul-bar, and from proprio-spinal fibres descending from the foremost segments of the neck. This is done — as the term “successive degeneration” implies — in a preliminary lesion by severing that part of the spinal cord it is desired to examine for the particular proprio-spinal fibres sought, from all the central nervous system lying farther forward. To determine the proprio-spinal fibres descending from the third and fourth thoracic segments, the first step is to transect the cord between the second and third thoracic segments. There then ensues throughout the length of the cord behind that transection degeneration of all the fibres that enter it from the brain, mid-brain, bulb, and cervical and first two thoracic segments. This heavy degeneration, after developing, reaches a maximum and gradually passes away, all the debris of the degenerated nerve-fibres being in time removed. For this a period of a year suffices in a dog. The spinal cord is then ripe for determination of the proprio-spinal fibres it is desired to examine. It becomes once more a clean slate on which a new degeneration can be written. The proprio-spinal fibres are revealed by making a transection between the fourth and fifth thoracic segments. Four weeks after this second lesion the proprio-spinal fibres descending from the third and fourth thoracic segments are degenerate. They can be studied (Fig. 17) throughout their course from the fourth thoracic segments backwards along the cord by any of the ordinary methods, such as the Marchi method, for studying degenerate fibres. Many of these proprio-spinal fibres pass from the shoulder segments to end in the hind-limb segments. The existence of the neuroglial scar, left by old degeneration that has cleared up, far from complicating the tracing of the new degeneration assists by forming a contrast background to it, the sharpness of which leaves nothing to be desired.

From the results obtained by this method the following reflex-chain can be inferred²⁵¹ as possible and probable for the scratch-reflex.

1. The receptive neurone (Fig. 13 B, ra) from the skin to the spinal gray matter of the corresponding spinal segment in the shoulder.

2. The long descending proprio-spinal neurone (Fig. 13 B, pa) from shoulder segment to the gray matter of the leg segments.

3. The motor neurone (Fig. 13 B, f c) from the spinal segment of the leg to a flexor muscle.

This chain thus consists of three neurones. It enters the gray matter twice; that is, it has two neuronic junctions, two synapses. It is a disynaptic arc.

In venturing to thus schematically express the construction of this arc as disynaptic, I am influenced by the desire to express its construction as simply as possible so far as consistent with the ascertained data of the case. I therefore omit from the scheme the possible Schalt-Zellen (v. Monakow) between ra and pa, and between pa and f c. Much of what I intend to express by “disynaptic” would be as clearly though not as concisely expressed by saying that gray matter is intercalated in the arc twice i. e., at two separate places. That, however, is not expressible by a single adjective, and since synapses occur so far as we know only in gray matter, disynaptic does include that idea. But it also implies somewhat more. I venture upon it in spite of the assumptions it includes because, in my opinion, much in those further assumptions seems justifiable and useful as a working hypothesis, and because it lays stress on the importance of the synapse in reflex conduction,—an importance which for reasons given before seems considerable.

The reflex-arc consists, therefore, of at least three neurones. It is convenient to have a term distinguishing the ultimate neurone f c from the rest of the arc. For reasons to be given later it may be spoken of as the final common path.³⁰⁰ The rest of the arc leading up to the final common path is conveniently termed the afferent arc.

The morphological components of this reflex mechanism include, in addition to the above neural elements, the muscle-fibres of the flexor muscle at one end, and possibly a receptive cutaneous organ at the other end probably in the root-sheath of a hair. Somewhere in this chain of structures the property of refractory phase has its seat, and refractory phase is a pivot on which the whole co-ordinating mechanism of this reflex turns.

In attempting to locate the seat of the refractory state, considerations that arise are the following. The muscle involved is one which neither when excited directly nor through its motor-nerve exhibits this refractory period. We can exclude the phenomenon, therefore, from it, and from its motor-nerve, and from the link between it and its motor-nerve,—the end-plate. Further, this refractory state is not exhibited when the motor neurones of this muscle are excited to activity by various other channels; for instance, by the afferent neurones coming in from the receptive organs of the leg itself, or by the palliospinal pyramidal neurones descending to them from the cortex of the cerebral hemisphere. We are free, therefore, to exclude the motor neurone f c supplying the flexor muscle itself as the source of the refractory state characterizing the scalptor-reflex. Again, the elicitation of the reflex typically (Fig. 14) by all the various above-mentioned forms of artificial (electrical) stimuli applied through a needle electrode inserted in the skin suggests that it is the commencement of the receptive nerve-fibres in the skin rather than any specialized cutaneous sense-organ therein that are the seat of stimulation when the reflex is thus artificially excited. We have no knowledge of the existence of a refractory phase of such duration as this as a property of any afferent fibre passing from the skin to the spinal cord. There is indeed conclusive evidence that the seat of this refractory phase lies neither in the skin nor in the afferent neurone itself. When the reflex is in progress under stimulation of the skin at one point, e. g., ra (Fig. 13, B), stimulation at some other, even remote, point also producing the reflex—as can be proved in various ways—does not break the rhythm of the reflex^{287, 300} or complicate it in any way. The reflex elicited from spot rβ may be initiated while that elicited from ra is in progress, and may then be carried on while A is allowed to cease, or vice versa; but in neither case is the rhythm of the reflex broken or reduplicated. And this although, from other evidence, we know that the flexor muscle and the motor neurone f c can respond rhythmically to stimuli with a rate of rhythm more than twice as high as that of the stimuli applied to produce the reflex. Or the reflex from a spot B may be introduced into the middle of one from a spot A, and although its reflex (B) impresses characters of its own as to amplitude, direction of the foot, etc., it does not reduplicate or break up the already existing rhythm. The result does not differ whether the individual stimuli of the two series of stimulations at spots A and B are alternate or not. Figures 18, 19, 20, 21 illustrate these points. The refractory phases obtaining in the reflex arising from spot A are respected by the stimuli delivered at B also and are not broken down by them.

There is evidently some part of the reflex mechanism which is common to impulses started both at A and at B. And this holds when spots A and B lie even ten centimetres apart. It is against what we know of the receptive neurones to suppose that there is between them any collateral nexus beyond their impingement directly or indirectly on other neurones more or less common to them both. The seat of the refractory phase seems therefore to lie somewhere central to the receptive neurones in the afferent arcs. The refractory phase induced in some element of the arc by the reflex from A extends to some element which is also concerned in the conduction of the reflex induced from B. This element must be some neurone common to the two arcs from A and B respectively. Neurone fc (Fig. 13), the final common path, is such an element. But neurone f c as tested by other reflexes, e. g., the flexion-reflex shows no such refractory period. The common mechanism sought for seems therefore to lie somewhere between fc and Ra, Rβ. It may well be that neurone pa is partly common to Ra and rβ, for these r neurones are well known to split intraspinally into headward and tailward stem-fibres, each carrying many collaterals, and probably by them connected with the gray matter in not only one spinal segment but in a series of segments. Collaterals from rβ as well as from ra may reach pa, therefore; and similarly with pβ.

The scratch-reflex has instructive points of likeness to that of the swimming-beat of Medusa. The arrangement of its response is quite like that of the muscular response of the swimming-bell of Medusa under stimulation of two points of the subumbrella or two of the marginal receptor organs. We can compare each lateral half of the saddle-shaped receptive area of the dog’s back in regard to the scratch-reflex quite strictly to the marginal surface of Rhizostoma in regard to its swimming-beat reflex.

In Medusa a second stimulus following close upon a first does not prolong the contraction;⁷² it finds the bell in refractory phase. The beat induced by the first stimulus has no second contraction fused with it in consequence of the second stimulus. The principles of the co-ordination thus obtained in the relatively simple swimming of Medusa seem as follows.

One condition of the co-ordination of the swimming-beat, as said above, is the unpolarized nature of the nerve-channels allowing free flow of nervous impulses in either direction from conductor to conductor along the nerve-net. Another condition is the continuity mediate or immediate of every conductor with every other. These conditions of the nervous system allow a single stimulus given at any single point to evoke a coordinate contraction of the whole musculature, to evoke, in short, a full and perfect swimming stroke or beat. But they do not insure that under a series of stimuli delivered in irregular and varied sequence and at various points in perhaps rapid succession a series of co-ordinate strokes or beats shall result.

Romanes⁷² showed that the receptor organs at the edge of the bell were the source of the natural beats of the bell, and that so long as even one of these remained the swimming-bell continued to beat spontaneously. There therefore exist in this case quite a number of points which all tend under natural circumstances to initiate the beats of the bell.

Suppose that shortly after a stimulus has occurred at A, and before the contraction induced by it has passed off, another stimulus is delivered at B, and then similarly at C. All these stimuli are, taken singly, similarly provocative of locomotion, and the only locomotive act of the creature is the systolic stroke of its bell. Each reaction is therefore aimed at incitement of locomotion only for a propulsive movement of the bell. But the two conditions, i. e., unpolarized conduction and end-to-end continuity obtaining in the nerve-net, though they insure that result for one stimulus, defeat it in the case of a series of stimuli quickly following either at one and the same point or at several,—and for the following reason. The series of stimuli would, were those two conditions all, merely immobilize the bell. The second stimulus following during the contraction excited by the first would be conducted, as was the former, to all the musculature of the movement, and would simply accentuate and prolong the systolic condition already in progress. But that would interrupt locomotion, not promote it. The contraction of the muscular disc to produce the stroke must be immediately preceded by diastole, enabling it to embrace the proper volume of water for re-expulsion.

An essential feature of the co-ordination is therefore due to alternation of the two converse phases of contraction and relaxation in the one muscle; these execute the locomotion of this simple animal. This necessary condition is insured, even under irregular series of stimuli, by refractory phase.

When the bell is replying or has just replied to a stimulus, it remains inexcitable to further stimuli for a period which outlasts its phase of contraction. This prevents disharmony occurring in the rhythmic movement under multiple stimulation.

It is on conditions like these governing Medusa’s swimming-bell that the co-ordination of the heart’s action is based, except that in the heart there is but one initiatory spot, prepotent permanently, as in Vertebrates, or temporarily, as in Tunicates. In Medusa presumably any one of various specialized points (border organs) becomes prepotent at different times, by reason of stimulation from the environment. In this Medusa more closely resembles the scratch-reflex, where any one of many points in a large receptive field becomes temporarily prepotent under stimulation, e. g., puncture by a flea or other parasite, and then initiates and leads a series of beats which can be prolonged or intensified by concurrent stimulation by parasites at other points, but cannot by their concurrence be upset as regards rhythm. In the action of scratching it is as necessary as in the swimming of Medusa, or the beating of the heart, that relaxation follow contraction.

Refractory phase is obviously an essential condition in the co-ordination of the scalptor-reflex. The scratching-reflex, in order to secure its aim, must evidently consist of a succession of movements repeated in the same direction, and intervening between the several members of that series there must be a complemental series of movements in the opposite direction. Whether these two series involve reflex contractions of two antagonistic muscle-groups respectively in alternate time, I would leave for the present. The muscle-groups or their reflex-arcs must show phases of refractory state during which stimuli cannot excite, alternating with phases in which such stimuli easily excite. Evidently this is fundamental for securing return to the initial position whence the next stroke shall start. The refractory phase secures this. By its extension through the whole series of arcs it prevents that confusion which would result were refractory phase in some of the arcs allowed to concur with excitatory phase in others.

But there is one significant difference between refractory state in the scratch-reflex and in the swimming mechanism of Medusa. In the latter, as in the heart, the refractory state is a property not relegated to a central nervous organ remote from the peripheral tissue in whose function it finds expression. It is located in intimate connection with the peripheral organ itself. From observations of Bethe it seems likely that refractory phase in Medusa is a function of the nerve-net. Magnus^268a has recently shown that the refractory phase of the beat of the isolated intestine is referable to the local nerve-plexus (Auerbach’s) lying in the gut wall. In these cases the refractory state seems to belong to the nervous elements, but to nervous elements diffused through the peripheral tissue. But in the scratch-reflex the site of the refractory state is central, intraspinal.

The centrality of seat of the refractory state of the scratch-reflex is significant of the difference of the conditions under which the scratch-reflex and the swimming of Medusa respectively go forward. In the case of the locomotor action of the swimming-bell of Medusa, we have a simple musculature which can execute practically but one movement. It is in fact a single muscle, that is to say, comparable with what in the more complex musculature of higher organisms—e. g., vertebrates—is regarded as a unit of musculature, a single muscle, such as the gastrocnemius, tibialis, etc., in the frog. Each and every receptor organ which under stimulation produces locomotion is therefore connected by nerve with that single muscle of locomotion, and when impelled by each or any of them, the muscle effects practically the same action as it does when impelled by any other of the sister receptor organs. The movement of locomotion which is provoked through each receptor is practically the same as that provoked through any of the rest. The mechanical organ in this case can perform but one movement, and its performance of that movement is, so to say, the one purpose demanded from it by each of all the receptor channels playing upon it.

But with the mechanical organ which the scratch-reflex employs the case is different. That organ is the hind limb, a complex structure built of parts, many of them spatially opposed, and able as a whole to execute movements of various kinds. Thus it can reflexly not only scratch, but stand, walk, run, or gallop, squat in defaecation, abduct and flex in micturition, etc. In the swimming-bell of Medusa there is no opportunity for antagonism between the motor end-results of the reflexes that employ it, save in respect to the possible confusion of successive contractions which would destroy the rhythmic pulse, and that confusion is avoided by refractory phase. The swimming-bell of Medusa is at the behest of but one type-reflex. The scratch-reflex possesses the same safeguard against destruction of its rhythmic character. But in the case of the scratch-reflex that reflex is but one of several reflexes that share in a condominium over the effector organ—the limb. It must therefore be possible for the scratch-reflex, taken as a whole, to be, as occasion demands, replaced in exercise of its use of the limb by other reflexes, and many of these do not require clonic action from the limb,—indeed, would be defeated by clonic action. It would not do, then, for the peripheral organ itself to be a clonic mechanism. The clonic mechanism must lie at some place where other kinds of reflex can preclude the clonic actuator from affecting the peripheral organ. Now such a place is obviously the central organ itself; for that organ is, as its name implies, a nodal point of meeting to which converge all the nervous arcs of the body, and among others all those which for their several ends have to employ the same mechanical organ as does the scratch-reflex itself. It is therefore only in accord with expectation that the seat of the refractory phase of the scratch-reflex lies where we traced it, in the central nervous organ itself, and somewhere between the motor neurone to the muscle and the receptive neurone from the skin. For it is upon the motor neurone that other arcs impinge.

The refractory state is obviously akin to a state of inhibition, and just as there are well-known examples in which the inhibitory state is peripheral (e. g., the heart), and others in which the inhibition is central, so undoubtedly phases of refractory state are in some instances peripheral, but also in numerous instances are central,—and this is so in certain reflex actions.

The reflexes of which refractory phase constitutes a prominent feature are those concerned with cyclic actions occurring in rhythmic series; such as the scratch-reflex, reflexes of swallowing and blinking, and probably the rhythmically recurring reflexes concerned in the stepping of the limbs.

Nothnagel (1870),⁴⁸ following up an observation by Setschenow,³¹ studied a periodic and rhythmic reflex of the crossed hind limb in the spinal frog. He found that if some days after the frog’s cord had been transected at the fourth vertebra, the central end of the sciatic is faradized, a rhythmic alternating flexion and extension of the opposite hind limb is evoked. He noted that no intensity of stimulation makes the rhythmic movement alter from clonic to tonic. A reflex of this kind is, I find, elicitable in the spinal dog’s hind leg by unipolar faradization of the opposite foot, and, as Nothnagel noted in the frog, no mere increase of intensity of stimulation converts its clonic character into maintained tonic. It is rhythmic (Fig. 22), and has a refractory period which no ordinary increase of intensity of stimulation suffices to break down. The frequency of its rhythm averages 2.3 per second, but varies somewhat in different observations. This is about twice as slow as the frequency of the scratch-reflex, which averages 4.5 per second. The average rhythm of the scratch-reflex is almost exactly the same as that ascertained by Gotch and Burch¹⁶⁴ for reflex-discharge from the electric cell of Malapterurus, but the rate in Malapterurus appears to vary much more than that of the scratch-reflex.

There is a peculiar brief extension-reflex of the dog’s hind leg which I term³⁰⁴ the “extensor-thrust.” Baglioni²⁸⁸ has more recently noted an analogous reflex in the frog. This reflex is elicited by mechanical stimuli applied to the planta. In the spinal dog, where well marked, it is often elicitable by even lightly stroking with the edge of a piece of paper the skin behind the plantar cushion. It is more certainly evoked by pushing the finger-tip between the plantar cushion and the toe-pads, especially when the hip and knee, not necessarily the ankle, are resting, passively flexed.

The “extensor-thrust” I have never succeeded in prolonging to a full half-second—none of my records exhibit it as long as that. Usually the duration as recorded is about a fifth of a second (Fig. 23.). Possibly its muscular contraction may be a simple twitch, though reflexly excited. Its muscular field involves the muscles of the “knee-jerk.” The myogram of the extensor thrust is shorter than that of tetanic contraction of the diaphragm (Head’s slip)¹¹⁸ caused by two successive stimuli to the phrenic nerve when the muscular responses of the two just fuse (Fig. 24).

Immediately after its elicitation this reflex, in my experience, remains in the spinal dog for nearly a second relatively inelicit able. Its reflex-arc exhibits after its phase of activity a refractory phase. The refractory phase is here far longer than that of the scalptor-reflex. It may last six times as long as the period of activity; thus the extensor-thrust may last only 170 σ while the succeeding refractory phase may endure a full second.

The extensor-thrust is probably an important element in the reflex mechanism of the dog’s locomotion.¹ One peculiarity it has as compared with other spinal reflexes of the limb is the considerable force which it exerts. In the locomotion of the animal it provides much of the propulsive power required. Bearing these points in mind, it is obvious that as an element in locomotion its repetition is required only at intervals considerably longer than the duration of the thrust itself, namely, at a particular phase of each successive step taken in the progression of the animal. After the extensor-thrust, the limb has to be given over to the flexor muscles in order, without touching the ground, to swing forward in preparation for the next step by the limb. It is reasonable to suppose that part of the means by which selective adaptation has secured this result is the evolution of the long refractory phase following the activity in the reflex-arc of the extensor-thrust. Zwaardemaker²⁷² has shown that the reflex movement of swallowing in the narcotized cat is followed by a refractory period,¹ lasting half a second or longer. This refractory state is central, for when the reflex swallow has been elicited by the superior laryngeal nerve of one side, the after-lasting refractory period holds good also for excitation of the opposite superior laryngeal nerve.

Variation of the external stimulus has comparatively little effect upon the length of the refractory period. But internal conditions, such as blood supply, fatigue, narcosis, etc., do influence it greatly. For reflexes which exhibit a refractory phase, a certain duration of that phase, subject to some variations, is characteristic. The duration of the phase varies considerably in different types of these reflexes.

It is clear that an essential part of many reflexes is a more or less prolonged refractory phase succeeding nervous discharge.

Refractory phase appears therefore at the one end and at the other of the animal scale as a factor of fundamental importance in the co-ordination of certain motile actions. In the lowly animal form (Medusa) it attaches locally to the neuro-muscular organ, and so also in the visceral and blood-vascular tubes of Vertebrates. But in higher forms (dog) refractory phase occurs as regards the taxis of the skeletal musculature, not in the peripheral neuro-muscular organ, but in the centres of the nervous system itself.