Discrete Threshold and Repetitive Responses in the Under ‘-Clamp”

I. TASAKI AND A. F. BAK From the Laboratory of , National Institute of Neurological Diseases and Blindness, National Institutes of Wealth, Bethesda, Maryland

ABSTRACT TASAKI, I. AND A. F. BAK. Discrete threshold and repetitive responses in the squid axon under ‘voltage-claw@ Am. J. Physiol. 193(z): 301-308. 1958.- By using the so-called voltage-clamp technique, records were obtained indicating the presence of undulatory or oscillatory membrane current under fixed potential difference between the internal and external elec- trodes. This oscillatory membrane current was observed only when rectangu- lar membrane depolarization was in the range between 15 and about 35 mv. The amplitude of the oscillatory membrane current was of the order of I ma/cm2. The oscillatory membrane current disappeared suddenly when the depolarizing clamping pulse was reduced below the ‘threshold.’ Various sources of artefact were examined, INTRODUCTION INCE Hodgkin, Huxley and Katz (I) membraneof the toad nerve fiber under volt- published the results of their voltage- age-clamp (3). Encouraged by the similarity Sclamp experiments, it has been widely between the toad nodal membrane and the accepted that, under the voltage-clamp condi- squid axon membrane under TEA, we ex- tions, the membrane current takes smooth, panded our observation on this discontinuous well-defined time courses,showing no sign of phenomenon to normal squid with an discontinuous threshold phenomena. During improved experimental setup. We found that the courseof a seriesof experiments designedto this phenomenoncould be demonstrated in the study the properties of the squid axon mem- majority of the fresh squid axons available at brane treated with tetraethylammonium Woods Hole. chloride (TEA), one of us (I. T.) working in The phenomenon in question can be stated collaboration with S. Hagiwara, encountered a as follows: when the ‘’ of strange phenomenon which appeared to the is raised above its resting conflict with the widely accepted view as to level by 15-25 mv and is kept at this level for a the behavior of the axon under voltage-clamp. period of about IO msec. or longer, the mem- The phenomenonobserved could be taken as brane current required to maintain the new signs of the existence of a discrete level undergoesrapid undulating or and of repetitive firing of responsesunder the ‘oscillatory’ changes which resemble the so-calledvoltage-clamp conditions. Since such phenomenon of repetitive firing of impulses a phenomenon was entirely unexpected and under ordinary experimental conditions. When appeared to be inconsistent with the basic the size of the clamping rectangular pulse is concept of voltage-clamping, no explicit reduced by small steps, this phenomenondis- statements were made of the phenomenonin appears suddenly at a certain intensity level, our previouspublications on the squid axon (2). indicating the existence of a ‘threshold.’ When Recently, we could demonstrate the the level of depolarization by the clamping existence of a similar phenomenonin the nodal pulse is about 40 mv or higher, the time course of the membrane current observed always Received for .publication November 7, 1957. resemblesthat published by the British in- 1 The work presented in this article was carried out at the Marine Biological Laboratory, Woods Hole, vestigators (I). Mass. In pursuing this phenomenon,we were fully 302 I. TASAKI AND A. F. BAK

used most frequently. The arrangement of diagram B is similar to that used by the n British investigators (I) and was used mainly in the early stage of the present investigation. The arrangement of diagram C is a combina- tion of the two, A and B. In the arrangement of diagram A, the internal electrode-set consisted of three silver wires of about 50~ in diameter. The potential- electrode, E,, had an exposed surface of 3-6

km mm in the middle of the current-electrodes. /////,II’// ///,II/// I/ EP ///~~pwqw~//, E p The main current-electrode, Em, had a 6 mm ’ f E; p fE; (sometimes 3 or 8 mm) long uninsulated FIG. I. Diagram A: experimental setup used for the surface, and the lateral current-electrode, El, experiments of figs. 2,3 and4. The drawings of the axon had two bare portions of 4 (sometimes3 or 8) and the internal electrodes are disproportionately short mm length adjacent to the bare portion of the and thick. The rectangular voltage pulse at the input of the high-gain differential amplifier, AZ, is used to main current-electrode. These internal silver clamp the membrane potential; the other rectangular wire electrodes were used both with and with- pulse is used to obtain records of ordinary action poten- out chloriding the surface electrolytically. The tials. Further details are in the text. Diagram B: ar- external potential-electrode, Ei, was a large rangement used for the experiment of fig. 7. Partitions, silver wire covered by a layer of gauze and P, are made with glass-plates (I mm thick) and with a small amount of Vaseline. Diagram C: arrangement agar gel. The ground electrode, which was used for the experiment of fig. 5. Partitions (made with similarly covered with gauze and agar gel, had Lucite plate and Vaseline) are watertight. a resistanceof about IO ohms to current pulses used in the present experiments. The sig- aware of the possibility that there might have nificance and the effect of the lateral current- been somekind of artefact in our experiments. electrode will be discussedunder RESULTS. We made therefore a number of control experi- The time course of the potential difference ments designedto explore such a possibility. between the two potential electrodes, E, and We try in the present paper to present the Ei, was recorded with a unity-gain differential results of these experiments as objectively as preamplifier, A 1 in figure I, diagram A (cf. possible. fig. 2 in ref. 2) ; the recording was as a rule A preliminary report of this work has been d.c.-coupled. The membrane current, 1%,was published elsewhere(4). determined by recording the IR-drop across METHODS resistance Ye which was generally between 2 and IO ohms. The high-gain differential The material, the experimental setup and the amplifier for negative feed-back, AZ in diagram procedures employed in the present experi- A, was a Tektronix amplifier of type I 12 (in- ments were similar to those described in a stead of type I 2 2 in the previous experiments), previous paper (2). Giant axons, taken from modified to give an output impedance of the squid available at Woods Hole, were used, about 250 ohms. This was achieved by adding in many caseswithout extensive cleaning of paired cathode-follower stages (two 6CL6 the surrounding tissues. In some cases,they tubes) at the output of the commercially were used after extensive cleaning under dark- available amplifier. The over-all voltage field illumination. The axon was mounted amplification, a, by this amplifier was ap- horizontally on a glass-plate of about 42 mm width, and a set of internal electrodes, con- proximately 2,300. The resistance connected sisting of two or three twisted silver wires was between the output of this amplifier and the inserted along the axis of the axon. current-electrodes, r in the diagram, was varied The arrangements of the internal and between 500 and 25,000 ohms. The smallest external electrodes employed in the present value of r/n (cf. p. 864 in ref. 2) available was experiments are illustrated by three diagrams therefore about 0.3 ohm. in-figure I. The arrangement of diagram A was As in the previous experiments (2), negative DISCRETE THRESHOLD AND REPETITIVE RESPONSES IN SQUID AXON 303 feed-back was accomplished by condenser- coupling. This did not give any limitation to our observation, since the duration of rec- tangular clamping pulses was in most cases shorter than 20 msec. and furthermore the time courseof the membranepotential, V, was always monitored by the second beam of a dual-beam oscillograph (DuMont type 322). Precautions were taken to eliminate all the causesfor a shift in the membrane potential resulting from closureof the manually operated key between the feed-back amplifier and the FIG. a. Records of membrane current associated with internal current-electrodes. The feed-back rectangular depolarization of a normal axon taken with circuit was closedimmediately before the start the arrangement of fig. IA. Dotted lines in records A to E show the potential difference between the internal of a sweepof the oscillograph beamsand was and external potential-electrodes controlled by the opened immediately after a photograph of the method of voltage-clamp. Record F was taken by potential and current traces was taken. A passing a short pulse of outward current through the period of at least IO seconds(often 30 sec. or membrane and later keeping the membrane current at more) was allowed before another clamping practically zero. Blanking, 0.1 msec. interval. The effec- tive length of the main current-electrode (Em), was 6 pulse was delivered. Most of the axons ex- mm, that of the lateral current-electrode 3 mm each amined showedno signof progressivedeteriora- and the axon diameter 4oo p. Temperature, 22°C. tion during the course of experiments which sometimeslasted more than 3 hours. inward surge which has slow rising and falling Most of the experiments described in this phases. However, we rarely obtained oscillo- paper were carried out at room temperature graph records similar to those reported by the which was between 20 and 23°C. British investigators. Two most common RESULTS examples of our records are reproduced in Membrane Currents Caused by Clamping figure 2 and figure 3, illustrating the difference Pulses Near Threshold. When the membrane between our results and those of the previous potential of a giant axon was suddenly raised investigators. above its resting level by a rectangular clamp- In our records an upward deflection of the ing pulse of 50-90 mv, a transient surge of potential (lower) trace representsa rise in the strong current flowing inward through the axoplasm potential referred to the potential of membranewas observed (seethe lower column the surrounding sea water; this recording polarity waschosen to show an ordinary action of fig. 2). The current intensity at the peak of the inward surge decreasedlinearly with the potential (record F in fig. 2) as a transient size of the clamping pulse in this range. When upward deflection. The polarity of the current the membrane potential was clamped at the (upper) trace is opposite to what was used by level of the peak potential of the action the British investigators; the reason for potential (observed without clamping), the adopting this new convention is to represent membranecurrent at the peak of the inward the behavior of an ohmic by a paral- surgewas close to zero. For negative clamping lelismbetween the deflections of the two traces. pulses, namely when the membrane potential With a pulse intensity of about 15 mv, or was clamped at levels below its resting level, slightly less, we generally obtained a weak the membrane behaved like a passive, ohmic inward membrane current (of the order of 50 (slightly nonlinear) resistor with a parallel pa/cm” or less) which smoothly changed into condenser.These findings agree perfectly with an outward membrane current. When the what has been reported by the British in- pulse intensity was gradually increased from vestigators. this level, there was, in almost all the ‘normal’ In the range of rectangular clamping voltage axons (i.e. in those producing action potentials between IO and 40 mv, Hodgkin et al. reported larger than IOO mv), a suddenappearance of a the appearance of a single small and smooth large inward membrane current (recordsB in 304 I. TASAKI AND A. F. BAR

changed gradually into the type of a smooth damping oscillation. When the clamping pulse reached the level of 40-50 mv above the , the multiple peaks in the inward membrane current disappeared and the records obtained resembled those published by the earlier investigators except for the fact that the maximum inward current observed was four to nine times as strong as those reported by Hodgkin et al. FIG. 3. Records showing the time course of the mem- In order to examine the existence of a brane current (upper truce) observed when the potential threshold and a sign of repetitive responsesin difference across the axon surface (lower trace) was clamped at various levels between rg and 30 mv above our axon under voltage-clamp further, the the resting potential. In record C 2 sweeps at slightly following control experiments were performed: different pulse intensities were superposed. Time a) investigation of the effects of imperfection markers, I msec. apart 22T. of voltage-clamping, b) the possibleeffect of the seriesresistance of the membrane, c) the effect fig. 2 and B and C in fig. 3). In other words, of the guard system and d) the effects of var- there was a discontinuous, true ‘threshold’ in ious chemicals. our experiments. The magnitude of this Effect of Feed-Back and Compensation.As discrete inward current varied to a consider- has been stated under METHODS, the feed-back able extent with the state of the axon; in amplifier used in the present seriesof experi- many cases it was of the order of I ma/cmz. The ments has a gain of 2300 and its output im- time interval between the beginning of the pedance was about 250 ohms. By controlling clamping pulse and the start of the discrete the resistancebetween the feed-back amplifier inward surge depended markedly upon the and the internal current electrodes (r in the size and the duration of the clamping pulse; at figure) between 500 and 25,000 ohms, we room temperature it was sometimes longer examined the effect of imperfection of voltage- than a few milliseconds. clamping. It was our repeated finding that the The first discrete surge of inward membrane variation in the amount of feed-back in the current was, as a rule, followed by several range just mentioned had very little or no similar inward surges. The configurations of clear effect upon the rhythm and the amplitude these inward surges varied in many cases from of the repetitive inward surges.An example of one surge to another, but in some cases (e.g. in such observations is presented in figure 4, top. record D in fig. 7) they appeared to be uniform. When the feed-back was the weakest within It was frequently seen that these individual the range mentioned above (left upper record), surges had a more-or-less abrupt beginning and there was a slight elevation in the membrane ending and consequently a succession of these potential at the time when the inward current surges resembled the pattern of repetitive reached a peak. Evidently, the feed-back was firing of impulses in an unclamped axon insufficient to control the membranepotential membrane. The interval between individual when the membrane resistance underwent a surges was close to I msec. (0.5-1.5 msec.) at marked reduction. When resistancer between room temperature (22T). In a series of the internal current-electrodes and the feed- experiments to be published later, it will be back amplifier wasreduced to about 3000ohms shown that this interval is prolonged markedly or less, the trace displaying the potential by a fall in temperature. difference between the internal potential- When the size of the clamping rectangular electrode and the external indifferent electrode pulse was raised from this threshold level took a perfectly rectangular time course. In all gradually, the latency of the first discrete (more than 30) axons examined, the repetitive- inward surge was decreased gradually, the ness in the membrane current remained amplitude of the first discrete surge was in- unaffected when the feed-back was increased creased gradually and the pattern of the un- to the highest value in the range mentioned dulating or oscillatory membrane current above. DISCRETE THRESHOLD AND REPETITIVE RESPONSES IN SQUID AXON 305

It has been postulated (I) that there is a CLAMPING IMPERFECT PERFECT COMPENSATED resistanceof about 5 ohm. cm2in serieswith the excitable membrane which consists of a variable resistancewith a parallel condenser. The small resistanceof the axoplasm between the internal electrodes and the surface mem- brane of the axon is undoubtedly connected in serieswith the membranecapacity. This series resistance should be smaller than about IO ohm.cm2 in our preparations, since the entire resistance between the internal and the I I I # m*ec external electrodes is between IO and 15 FIG.4. Upper column: records of the current through ohm.cm2 in most of our preparations (cf. the main current-electrode (r, in fig. IA) taken when p. 873 in ref. 2). We made a seriesof experi- the feed-back for clamping the ‘membrane potential’ ments in which the effect of this series re- was insufficient (left), when it was sufficient (middle) and when the ‘series resistance’ of the membrane was sistance was compensated by an external compensated (riglzt). Lower column: Current through circuit. the main current-electrode (left), that through the The principle of the method of compen- lateral current-electrode (middle) and the difference sating the seriesresistance we employed is as between the 2 currents (right). The rectangular clamp- ing pulse was approximately 22 mv in all the records in follows: In diagram B of figure I, the area of this figure. The experimental arrangement was the the axon membrane in the middle pool is same as that for fig. 3. close to 0.1 cm2. If one assumesthat the clamped portion of the axon membrane is ohms (about 5 ohm.cm2) has actually been uniform, the series resistance of 5 ohm.cm2 compensated. This type of compensation was should then give rise to a potential drop of 50 found to reduce the amplitude of individual ohms times the current passing through the inward surges, reflecting the falls in the membranein the middle pool. When resistance membrane potential at the peaks of inward r, connected to the middle pool is 5 ohms, the surges.However, the pattern of the repetitive- IR-drop acrossthe seriesresistance should be ness remained essentially unaffected by the IO times as large as that acrossr,,. We ampli- procedure. fied the potential drop acrossr, IOOO times Effect of the Guard System. When the with a differential amplifier (Tektronix no. 122) membrane potential is clamped along a and, using a potential divider which reduced rectangular time course, the internal current- the potential drop by a factor of IOO, the signal electrode carries a current that flows near the was superposed upon the rectangular pulse end of the current-electrode (where voltage- used to clamp the membrane potential. The clamping is imperfect) as well as the current membranewas thus clamped by the amplified which is actually used to clamp the voltage. IR-drop across r,,, superposed upon a rec- The arrangement of figure IB, was used tangular pulse. By varying the value of r, and previously (I, 2) to eliminate the current also the potential divider, it was possible to flowing near the end of the current-electrode. compensatefor the seriesresistance of about In this arrangement, the fluid medium sur- 7 ohm-cm2or less. This method of compensa- rounding the axon is divided into three separate tion was applied also to the arrangement of pools by two partitions. If the potentials of diagram A in figure I. the three pools are maintained at the same In the right, upper record of figure 4, it is level during voltage-clamping, the current seen that the membrane potential goes down at the moment when there is a strong inward that flows through the middle electrode should surge in the current trace. (Note that the sign represent the membrane current in the uni- of potential variation is opposite in the left formly clamped portion of the axon. However, upper record.) By separating the recorded a strong current required to clamp the mem- membrane potential into an IR-drop super- brane potential tends to set up potential posedupon the rectangular pulse, it is found differences among the pools which can cause that the seriesresistance of approximately 65 flows of current through the leakage resistance 306 I. TASAKI AND A. F. BAK

use of the methods mentioned above to test whether or not the multiple inward surges described above arise in the portion of the axon near the terminal of the internal current- electrode. In all casesthe result of our observa- tion supported the view that these inward FIG. 5. Effect of urethane upon the repetitiveness of surgesarise in the ‘clamped’ regionsof the axon the axon membrane under ‘voltage-clamp.’ The ar- rather than from the terminal portion of the rangement of fig. IC was used, in which the middle clamped axon. An example of the experiments pool was 4.5 mm wide, partition about 2.5 mm wide. along this line is presented in the lower column the effective length of electrode Em 5 mm, that of Et 5 mm each. The lateral pools were filled with .?Y JW- of figure 4. In this example, the effective length thane-sea water. The fluid in the middle pool ia”, r%st of the main current-electrode (exposed for 6 normal sea water (left), next urethane-sea water mm) was equal to that of the lateral current- (middle) and then normal sea water again (left). Cali- electrode (3 mm on both sides).The amplitude bration: 50 mv for the potential (lower) trace and I ma/cm2 for the current (upper) trace. Time markers, of the oscillatory membranecurrent traversing I msec. apart. the resistance (rm) connected to the main current-electrode was slightly larger than that across the partitions. This can cause an error traversing the lateral electrode. Consequently, in the measurementof the membrane current. the difference between the two currents showed Nevertheless, we sometimes employed this a small oscillation which was approximately in method in the present seriesof investigations. phase with the current flowing through the The secondmethod employed in the present main current-electrode. Effect of Narcosis. The effects of alcohol and study is illustrated by diagram A in figure I, in which an ‘internal guard system’ is adopted. If urethane upon the current-voltage relationship the potential difference between the two in- of the axon membrane under voltage-clamp ternal current-electrodes is maintained at a will be reported fully elsewhere.The effects of low level (in practice within a fraction of I these narcotics upon the repetitive inward mv), the potential drops across resistor r, surges can be summarized as follows: in a deeply narcotized (but not totally inexcitable) (2-10 ohms) should give a faithful index of the currents carried by the main current-electrode axon, the of which was to the membrane. The effective length of the completely graded, there was no sign of dis- main (middle) current-electrode used was continuous phenomenon or repetitive mem- between 3 and 6 mm, and that of the lateral brane currents. Under light narcosis, various one was 3-8 mm on each side of the main degreesof transition between the behavior of electrode. the normal axon and that of a deeply nar- The third guard system employed is shown cotized axon were observed. by figure IC, which is a combination of the two The records furnished in figure 5 were ob- methods mentioned above. With this third tained by using the arrangement of figure IC. method, it is possibleto compare the current The portions of the axon in the lateral pools recorded with the internal current-electrode were rendered inexcitable with a 3 % urethane- with that observed with the external current- sea water solution. Recordings from both the electrode. In combination with these guard internal and external main current-electrodes systems,we employed the method of narcotiz- indicated that there were repetitive inward ing the portion of the axon in the lateral pools surgesin the excitable membranein the middle with a 3% urethane-seawater solution. This pool. In the external lateral current-electrode, narcosisof the axon in the lateral pools tends there was a weak oscillatory current which to increase the leakage current across the had an amplitude of about one-fifth of that in partitions, but the inward currents observed the main current-electrodes; we attributed this under these conditions can be regarded as a to a spread of currents across the partitions. sign of physiological activity taking place in When the sea water in the middle pool was the axon in the middle pool. replaced with the 3% urethane solution, the We made a number of experiments by the membrane became nonrepetitive within I DISCRETE THRESHOLD AND REPETITIVE RESPONSES IN SQUID AXON 307 minute; in other words, the axon acquired a NORMAL character to respond to rectangular clamping pulses with single surges of inward current, the amplitude of which varied continuously with the pulse intensity. Approximately 1.5 minutes after introduction of the narcotic into the middle pool, the fluid in the middle pool was replaced with normal sea water. As can be seen in the record, there was a partial recovery in the repetitiveness of the axon membrane I % URETHANE under voltage-clamp. The effect of short narcosis of the axon in the middle pool was reversible and the observation could be repeated several times. The records in figure 6 show the gradual transition of the axon membrane from the FIG. 6. Loss of the discreteness in the membrane normal state to the narcotized. These records action potential (left) and in the membrane current were obtained with the arrangement of diagram under ‘voltage-clamp’ (3 records on right) by narcosis. A in figure I. When the axon was in normal sea water, the action potential of the axon (recorded under current-clamp) was highly all-or-none; that is, the jump in the membrane potential from the subthreshold level to the level of a full-grown response was large and discontinuous. Under deep narcosis, the action potential of the axon became graded; that is, the peak value of the action potential varied continuously with the intensity of the applied current pulse. Under these conditions, little or no discrete inward membrane current was observed when the potential difference be- tween the internal and external electrodes was FIG. 7. Action potentials of an axon treated with TEA (records A and B) and the membrane current altered along rectangular time courses. The under ‘voltage-clamp’ of the same axon (records C to P). effect of narcosis upon these properties of the The arrangement of fig. rB was used. Time markers, 5 axon was nearly perfectly reversible. msec. apart. Effect of TEA. As has been stated in the introduction, the repetitiveness under voltage- feed-back was not very strong in these experi- clamp attracted our attention first in axons ments, the sign of repetitiveness was clear. It treated with TEA. Since we did not include is interesting to note that the repetitive inward the information as to this property of the axon surges in these records resemble those observed in our previous paper (2)) we make here a brief in the nodal membrane of the toad (3) in its description of these early observations. characteristic that the membrane current was The records presented in figure 7 were taken almost purely inward-directed during the from an axon into which an isotonic solution whole period of voltage-clamping. of TEA had been injected uniformly from one end of the axon to the other. The technique of DISCUSSION voltage-clamping in this experiment was The range of the membrane depolarization described earlier (2). The properties of the in which the repetitive inward currents under prolonged action potential (records il and B) voltage-clamp can be demonstrated is not so and the voltage-clamp behavior for large narrow (roughly between 15 and 35 mv) as to depolarizing pulses (record F) were also fully be easily overlooked during a standard voltage- discussed in the earlier paper. Although the clamp measurement. There are three possible 308 I. TASAKI AND A. F. BAK

explanations for the discrepancy between our tend to alter the axon into a less repetitive observations and those made by the British (damaged) state. A constant current which investigators. They are : a ) There is in our arises from the feed-back amplifier when an experiments some unknow n source of artefact, attempt is made to clamp the membrane b) squid axons available in England (Loligo potential continuously above or below the forbesi) are not as repetitive under voltage- resting potential can be a causeof a progressive clamp as those we used in Woods Hole (L. damageof the axon. The degreeof cleaning the pealii), or c) there is a difference between our axon or of chloriding the surface of the internal technique of voltage-clamping and that of the silver electrodesdid not have a clear effect in British investigators. our experiments. During the course of preparation of the We have frequently observed that a local present paper, we showed our photographic surgical injury (e.g. compressionor traction) records to Drs. Grundfest, Shanes, Freygang altered the property of the axon locally and and their associates and asked if they could rendered the membrane less repetitive under demonstrate the repetitive phenomenon with ‘voltage-clamp ; in these experiments, when their voltage-clamp apparatus. It was our the clamping electrodes were shifted to the pleasant surprise that their instrument which neighboring uninjured part of the axon, normal is similar to that used by the British investiga- repetitive inward surges were generally ob- tors gave results similar to those presented in served. It was our consistent observation that, figures 2 and 3 of the present paper. Although whenever the discretenessin the membrane this fact does not completely exclude the current under ‘voltage-clamp’ was lost, the possibility that there might still be some axon was not capable of developing an action artefact in our experiments, the difference in potential in a clean all-or-none manner under the feed-back apparatus appears to be com- current-clamp. pletely excluded. In the following paper, evidence will be It is not possible to discuss the second pos- presented showing that the discretenessin the sible explanation until we hear more about the membranecurrent under ‘voltage-clamp’ arises squid available in England. from patchy excitation of the axon membrane. The third possible explanation for the ACKNOWLEDGMENTS discrepancy between our results and those We express our thanks to Drs. H. Grundfest, A. M. published by the British investigators is to Shanes and W. H. Freygang for their valuable help and advice. We are indebted to Dr. C. S. Spyropoulos for attribute the discrepancy to a difference in the his help in the experiment. clamping technique. We took a precaution not to repeat our clamping pulses at a high rate; REFERENCES an interval of at least 10-30 secondswas al- I. HODGKIN, A. L., A. F. HUXLEY AND B. KATZ. J. lowed for the axon to recover from the passage Physiol. 116: 424, 1952. of a strong current during the period of volt- 2. TASAKI, I. AND S. HAGIWARA. J. Gen. Physiol. 40: 859, =957a age-clamping. It was our impression that 3. TASAKI, I. AND A. F. BAK. J. Newophysiol. In press. clamping pulses repeated at a rate of I/set. 4. TASAKT, I. AND A. BAK. Science 126: 696, 1958.