Motor unit recruitment and the gradation of muscle force.

The force output of a typical skeletal muscle can be modulated over an enormous range, typically more than ten thousandfold. It is generally stated that this modulation of force output is accomplished by a combination of rate coding of individual motor units and recruitment of more or fewer motor units. These processes are so well known and so generally accepted that less thought is given them than they deserve. The purpose of this article is to examine some of the strengths and limitations of these processes. I will suggest some answers to such questions as: Why is it functionally useful to subdivide a muscle into motor units? Are motor units of very different types needed? is a fixed recruitment order advantageous, and is this always so?

These questions have been discussed before, but the answers given have not always been the same. It may be useful to summarize recent findings and to show how the study of motor units in the laboratory deepens our understanding of how a muscle functions, and may produce results useful to the practicing physical therapist.

Not all muscles are subdivided into motor units, nor is rate coding a possible way of modulating force in all muscles. The best example of a muscle in which force modulation is not possible either by rate coding or by recruitment of motor units is the mammalian heart. The heart muscle is an extremely slow, fatigue-resistant muscle, much more so than the much-studied soleus muscle of the cat. The action potential of ventricular muscle lasts about 300 milliseconds, and a twitch lasts about as long. Because the muscle membrane is refractory during the action potential, it is difficult or impossible to stimulate heart muscle until the twitch is nearly past. A fused tetanus cannot be produced, and although a partly fused tetanus may be generated, the necessary stimulus frequency is too slow to produce much force increase. The heart pumps with a series of twitches of the individual muscle fibers. This process is described in textbooks on human physiology,[1,2] and particularly good accounts may be found in the books edited by Mountcastle[3] and by Schmidt and Thews.[4]

Because heart muscle cells are electrically coupled, an action potential produced in one is conducted to its neighbors and excites them. Contraction proceeds over the muscle in a wave, involving all the muscle cells in a sequence determined by the current path and with a speed determined by the conduction velocity of the coupled fibers. Heart muscle is said to act as a functional unit, a syncytium. Stimulate one part of it, and the whole muscle contracts. There is no organization into motor units, and the recruitment of only a part of the muscle is not possible, at least in the normal heart. This way of controlling a muscle may seem strange to one who is accustomed to thinking mainly of skeletal muscles, but it works very well for the heart.

Skeletal muscles are called on to perform a very different series of tasks. The work of the heart changes from a minimum when pumping blood for a person at rest to about 10 times the work pumping blood for the same person working at his or her aerobic maximum. The working range of a skeletal muscle is several orders of magnitude larger. The heart accomplishes its task with a series of twitches. A skeletal muscle may produce an impulsive force, a steady tetanic contraction, or any of a variety of time-varying forces in between. It is not surprising that skeletal muscle is designed in a different way to produce this far greater variety of tasks. On the other hand, the heart must carry out its stereotyped task without interruption for a lifetime, whereas even the most fatigue-resistant skeletal muscle is granted periods of rest. This is also reflected in the muscle's design. We may say that a muscle divided into motor units offers the advantage that a greater or lesser part of the muscle may be activated, showing a range of force outputs limited largely by the number of motor units.

Motor Unit "Types"

Henneman and co-workers[5,6] were the first to examine the mechanical properties of individual motor units (reviewed by Henneman and Mendell[7]) and to show that motor units have very different speeds and strengths of contraction. At the same time, histological evidence[8] showed that three distinct types of muscle fibers could be identified, distinguishable by their oxidative metabolic capabilities, fiber diameters, and other properties. Henneman and co-workers inferred that all the muscle fibers of a motor unit were of the same histological type; this fact was directly demonstrated 3 years later by Edstrom and Kugelberg.[9]

At about this time (1967), Burke and co-workers (reviewed by Burke[10]) set about defining mechanical criteria that would separate motor units into three distinct types matching those found histologically. They succeeded in separating motor units into fast and slow, fatigable and fatigue-resistant units in such a way that three motor unit types emerged according to these mechanical criteria. A stubbornly unclassifiable motor unit type, with a fast twitch and an intermediate fatigue resistance, proved to appear consistently in cat hind-limb muscles and was designated as a fourth type (FI) in studies after 1974. The three basic motor unit types could be shown to be uniquely identifiable according to the metabolic pathways they used to produce force (reviewed by Burke[10]). The histologically elusive intermnediate unit was later identified by McDonagh and co-workers.[11] The names and correspondence of these motor unit types are

Fast, fatigable (FF) = fast, glycolytic

(FG)

Fast, intermediate in fatigability (FI, F

[int]) = FI (McDonagh et al[11])

Fast, fatigue resistant (FR) = fast, oxidative

glycolytic (FOG)

Slow (S) = slow, oxidative (SO), utilizing

glycogen and fat for energy

Most of the early studies were performed on the triceps surae group of muscles in the cat; in particular, the medial gastrocnemius (MG) and soleus muscles were studied in great detail. The soleus muscle turned out to be an unusual muscle, composed exclusively of slow motor units. The MG muscle was found to be a mixed muscle and can be regarded as typical. Many other muscles had histological staining patterns that showed they also possessed three muscle fiber types, but for a variety of technical reasons, mechanical properties of their motor units were not examined immediately. As a consequence, the tests of speed and fatigue resistance so suitable for motor units of the MG muscle became standard tests. This is unfortunate, because some muscles, particularly in humans, differ so much from the cat MG muscle in speed and fatigability that the "standard tests" are of little use.

The histochemical classifications have also been questioned. Many histochemical studies, particularly those involving human muscles, divide motor units into the three types (I, IIa, and IIb),[12-14] whereas a type intermediate between types I and IIa, called IIc, is occasionally mentioned.[14-16] The following equivalence is often made in the histochemical domain alone:

SO = I = B,

FOG = IIa = C

FG = IIb = A where the A, B, C classification used by Henneman and Olson in their original work8 matches the SO, FOG, FG classification quite well, but is less commonly used today. A number of other classifications have been used,[17] which will not be mentioned here.

Making such a table of equivalences is risky because different histochemical methods are used to establish the classification schemes, and they need not be truly equivalent. The reason is worth examining briefly because data from the literature are often misinterpreted. A detailed discourse of histochemical classification schemes, however, could easily form a review on its own and is best left to an expert in that field. It is beyond the scope of this review. Brooke and Kaiser attempted to make a table of equivalent classifications and abandoned the idea with the words

.. although it is true that in a given

animal or even another muscle in the

same animal such a correlation can be

made, when another animal or even

another muscle in the same animal is

considered, the correlations are different.

To take a very simple example, the

type IIa fiber in a rat gastrocnemius is

a red or C fiber; however, in the human

biceps, the the IIa fiber is an

intermediate or B fiber.[17(p135)]

The problem arises because a muscle fiber is a complex system, the function of which can be tested in many ways that are not equivalent. Henneman and co-workers, in their original work (this has been reviewed in detail[7], stained the mitochondria in muscle fibers; they described A fibers with few mitochondria (white, FG), C fibers that stained richly for mitochondria (red, FOG), and an intermediate type B (SO). Peter et al[18] and later Burke[10] used stains that demonstrated the oxidative capacity of the myofibrils of muscle fibers to identify them as SO, FOG, or FG fibers. Often, several stains were used on successive sections of the same fiber.

The methods used to classify fibers as I, lIa, or IIb are quite different.[12] The muscle fibers are incubated at a known pH (preincubation). They are then stained with chemicals that show the ability of the muscle fiber to utilize adenosine triphosphate, an ability that may have been destroyed by the preincubation. If the adenosine triphosphatase (ATPase) activity remains intact in the pH range of 3.9 to 10.8, it is considered acid stable and identifies the fiber as type I. Group II fibers are acid labile and retain their ATPase activity after preincubation in a bath within a pH range of 4.5 to 10.8 (type IIb) or 4.9 to 10.8 (type IIa). These values are for human biceps or vastus lateralis muscles[l2] and may differ in other muscles or in animals.[14,15,17] It is now known that the pH stability depends on the structure of a heavy chain of the myosin molecule; the significance of this for myosin ATPase activity is not fully understood.[19]

In humans, types IIa and IIb also overlap greatly in their aerobic capacity (ie, in the ability to identify them as FOG or FG).[14,15] The classical IIa fiber of rodents, with its strong oxidative and glycolytic capacity, is quite different from the human IIa fiber, which stains only moderately for the enzymes of oxidative and glycolytic activity.[19] This difference should not be surprising; there is no compelling reason to suppose that the acid stability of the enzyme ATPase in a muscle fiber (I, IIa, IIb), the oxidative and glycolytic enzyme content of the fiber (SO, FOG, FG), and the number of mitochondria the fiber contains (A, B, C) should all be perfectly correlated. Some histochemists have pursued this idea, performing multiple staining tests and dividing muscle fiber into as many as nine types.[20,21]

Motor units in different muscles may differ widely, even when they carry the same names. The Table lists the motor unit populations of several muscles and gives some properties of the motor unit types. All data listed were obtained from muscles of the cat. The list is neither representative nor complete. Instead, the muscles have been selected deliberately because their motor unit compositions, or the properties of their motor units, are very different. Because of its familiarity, the classification of Burke and co-workers has been retained in most publications, even when the original definitions do not apply. It is important to realize that a fast motor unit in one muscle may be very different from a fast motor unit in another muscle of the same animal. The properties of motor units of the same muscle in different animals may also differ widely. Figure 1 shows model twitches of FF and S motor units in the MG muscle. The twitches have been generated with a simple mathematical equation but represent the typical shape of a twitch. In the upper panel, forces are shown to scale; in the lower panel, they are normalized so that speed differences are easier to compare. In any one muscle, it is fairly easy to identify three or four district motor unit types by differences in speed, strength, and fatigability. It is risky to extrapolate and assume that the properties of motor units with the same names in another muscle are the same.

Figure 2 makes this point by showing two FF units from different muscles, the flexor carpi radialis (FCR) and the diaphragm. Note that the FF unit of the FCR muscle is more powerful and faster than the corresponding FF unit in the diaphragmatic muscle.

Several interesting features of the relationship among motor units may be derived from the Table. In any muscle, FF motor units are the strongest, usually two to three times as strong as FR units (FI units tend to be an ambiguous group and, as discussed previously, lie in a position intermediate between FF and FR units). Note that the diaphragm is an exception in that FR units produce 75% as much force as FF units, on average. Measurements were made from intact diaphragms and from portions of the diaphragm detached at the costal margin. Only data obtained from the latter experiments were reported, because the tetanic force of all motor unit types was about 25% greater when the portion of the diaphragm, was detached from the ribs. Type S units are, as a rule, less than half as strong as FR units. The FCR muscle represents an extreme motor unit relationship. The average type S unit is only one fifth as strong as the average FR unit. Note also that the muscle is composed of 40% of these type S units. Thus, this large population of weak units produces only 8% of the muscle's maximum force.[22]

Twitch contraction times of type S units tend to be about twice as long as those of FF units. An exception is the extensor digitorum longus (EDL) muscle, in which all motor units are remarkably similar in speed.[23] Notice the differences in the motor unit populations of the diaphragm and of the peroneus longus muscle in this regard (Table). In both muscles, average twitch contraction times span a range of two to one. Yet the type S motor units of the peroneus longus muscle, with a mean twitch contraction time of 30.9 milliseconds, are actually faster than the average FF unit of the diaphragm, with a mean twitch contraction time of 34 milliseconds. These findings emphasize the statement made earlier that the tetrapartite motor unit classification works well within one muscle but should not be used as a global classification.

Finally, the percentages of the different motor unit types making up any muscle differ greatly. The classic example, not listed in the Table, is the cat soleus muscle, which consists exclusively of type S motor units. This homogeneity is true for the cat[6,8] and guinea pig,[24] but not for most other animals. For example, in the rat[25] and in humam,[26] the soleus muscle is a mixed muscle in which type S motor units predominate. The FCR muscle in the cat is composed of almost half type S motor units,[22] whereas the EDL muscle contains a mere 6% of these units.[23]

The muscles that move the eye are a world unto themselves (reviewed by Goldberg[27]). It is difficult to fit their motor units into the usual tetrapartite scheme. First, these muscles contain a population of nontwitch (NT) motor units, which only produce force when stimulated repetitively at rather high frequencies. Second, it has recently been shown that these muscles contain a population of slow, fatigable (SF) motor units.[28] It should be mentioned that all extraocular motor units are at least an order of magnitude faster and weaker than those of limb muscles; therefore, the tests by which speed and fatigability are measured must be strongly modified.[28]

Although it is convenient to speak of motor unit types, this does not mean that their properties are discrete and nonoverlapping. In any muscle, it is possible to find type S motor units that are stronger than some type FR motor units of the same muscle; the same is often true for speed. Although the Table may appear to list discrete properties of the various motor unit types, overlaps among most properties occur. The exception is usually fatigability; FF and FR motor units are readily distinguished according to fatigability using the test developed by Burke.[10] Of all the tests of motor unit properties, the results of this test are the easiest to interpret. Nevertheless, even this test has had to be modified, albeit with some controversy, to accommodate unusually fast muscles such as the FCR, peroneus longus, and extraocular muscles.[22,28,29]

By almost all measures then, motor units are most readily arranged in a continuum, and certainly they are brought into activity not according to some type or grouping, but smoothly in a continuous order.[7] A classification into "types" serves as a useful shorthand to identify extreme properties. Such classification shows the differences among motor unit properties such as speed, strength, metabolic pathways used to produce energy, and fatigability in a clear, readily understandable way. There are gradations in all motor unit properties, however, and all gradations show a continuum from one extreme to the other. It is this spectrum of properties that is taken advantage of when motor units are recruited to grade muscle force.

There has been considerable difficulty in classifying human motor units according to mechanical properties, in part because they cannot be isolated by the invasive techniques used in animal experiments. Human motor units are usually studied by microstimulation of fine nerve terminals in the muscle or by percutaneous nerve stimulation,[30-32] or by spike-triggered averaging.[33] A weak stimulus delivered percutaneously or through an electrode inserted into a muscle may trigger action potentials in a single, axon or axon branch. The spike propagates into all the branches of the axon, activating a single motor unit. Spike-triggered averaging is somewhat more complex. A fine-wire electromyographic (EMG) electrode in the muscle is used to identify the action potentials of a single motor unit, which is only possible at low to intermediate force output. It is assumed that the action potential can be reliably identified and isolated. The action potential is used to trigger a computer, which measures the force output of the same muscle. The computer stores the muscle force as a function of time for a short period (eg, 300 milliseconds) after the trigger pulse and adds up successive force records. Because the twitch from the motor unit in which the EMG spike is used as the trigger always occurs at the same time relative to the trigger, successive twitches are added in the computer. The twitches of other units occur randomly in relation to the trigger and tend to be smoothed out. in this way, the twitch of one unit may be selected.

In the first method, the number of motor units it is possible to isolate is small, and, in the second method, serious errors in measurement are possible.[34] Because, even at low discharge rates, motor units produce partially fused contractions, the time course of individual twitches is distorted and their amplitude is underestimated. Additionally, the spike-triggered averaging method is based on the assumption that only the signal of interest occurs in a fixed time relation to the trigger; synchronous discharges of motor units would produce artifacts.

These problems have been largely circumvented in a recent study on thenar muscles by a group working in the laboratory of Johansson in Sweden.[34-36] Fine tungsten electrodes were inserted into single axons of the median nerve to allow the stimulation of single motor units in the thenar muscles. Tetanic force ranged from 28 to 174 mN (2.9-17.7 g). Twitch contraction times ranged from 35 to 80 milliseconds. In contrast to the findings of most animal experiments, there was no correlation between strength and speed of motor units.

That is, there was no tendency for the strongest motor units to be the fastest, as is the case for the motor units shown in the Table, nor was it possible to divide the units studied into types in the conventional way.

Motor Unit Populations and Muscle Function

We have seen that the percentage of each motor unit type in a muscle differs widely from muscle to muscle. This difference is related to the task the muscle is most commonly required to perform. The vast majority of muscles are mixed, being composed of three or four motor unit, and hence muscle fiber, types. The cat soleus muscle is perhaps the best-known exception, being composed solely of type S motor units. The vastus intermedius muscle is also composed predominantly or exclusively of type SO muscle fibers,[24,37] although its motor units have not been studied, to my knowledge. At the other extreme are the tensor fasciae latae and gluteofemoralis muscles, which are composed almost exclusively of type FG muscle fibers[24] cited by McDonagh et al[37]).

The relation between fiber composition and function is most readily seen in the extensive study of the muscles of birds by George and Berger.[38] I know of no other example in which the same muscle has been studied in such a wide variety of species and shown to exhibit such a variety of fiber compositions. Although the relevance of bird flight muscles to physical therapy in humans may not be immediately apparent, this digression is presented for two reasons. First, it is intended to impress on the reader that the same muscle may have entirely different properties in different species. The "typical" cat MG muscle may tell us little about properties of that muscle in humans. Similarly, humans trained to perform specialized tasks, in a sport, for example, or subjected to movement constraints, as after a stroke, show a remarkable diversity in properties of the same muscle.[14,26,39,40] Second, a particular muscle is marvelously adapted to its particular function. This is most easily seen in the variety of flight muscles adapted to a variety of styles of flight. Flight is a form of locomotion with severe constraints produced by the laws of aerodynamics, so flight muscles must represent an extremely specialized design.

Only the muscle fiber types and their percentages are presented in the monograph by George and Berger,[38] and these fiber types are referred to as red, white, and intermediate. It is safe to assume that these fiber types correspond approximately to FOG, FG, and SO fibers, respectively, of the modern nomenclature. The authors describe the fiber compositions of the pectoral muscles, the flight muscles of birds of several classes.

Domestic fowl have pectoral muscles in which FG fibers predominate. This predominance is consistent with the flight behavior of chickens, a sprint-like activity with little endurance. Ducks and geese, by contrast, have pectoral muscles consisting predominantly of FOG fibers. This predominance is what one might expect of a migratory bird, which must be an endurance flier and yet engages in rapid wing movements. Pigeons and doves have pectoral muscles consisting of a mixture of FOG and FG fibers. Several other birds of unrelated types (eg, the cattle egret) have muscles of like composition. These birds may be considered to be sprinting fliers, but spend far more time in the air and are better fliers than fowl.

A group of birds that spends much time in the air would be expected to have flight muscles designed for endurance. Pectoral muscles composed exclusively of oxidative muscle fibers (FOG and SO) are found in a large variety of birds, including swallows and swifts, crows, red-winged blackbirds, and robins, to name a few. Hummingbirds, as might be expected, have pectoral muscles composed exclusively of FOG fibers. To my knowledge, there is no muscle composed only of FOG fibers in a mammal. Finally, several types of birds have pectoral muscles composed exclusively of SO fibers, like the soleus muscle of the cat. Such a muscle would be designed to produce steady force without fatigue for long periods of time, but would not be designed for rapid movement. This type of muscle is found in soaring birds such as kites and vultures. Although soaring is clearly locomotion, it can be argued that a soaring bird maintains a posture more than it engages in a locomotory movement.

The idea that muscle structure is exquisitely adapted to function is readily carried over to mammals. In the cat, the ankle flexor tibialis anterior (TA) muscle is composed pre-dominantly of FG fibers[23] (Table). The cat stands on its toes and uses this muscle for locomotion, but not for posture. In humans, this muscle is composed of 65% to 80% SO fibers,[26] and it is used to maintain our upright posture. We may conclude that there is no "typical" muscle; the fiber composition of a muscle depends on the use to which it is put by the particular animal, or human, it serves. Humans seem to vary in their physical fitness more than animals, and so show a wider range in the composition of their muscle fiber types than do most animals. This is borne out by studies[26,40] in which examination of the same muscles in autopsy or biopsy samples from numerous subjects shows a remarkable variability.

It is not yet known what determines the fiber type composition of a motor unit or of a muscle. The use to which it is put undoubtedly plays a role. It is likely that fiber type is determined in part genetically. Hoh[41] has suggested that myoblasts exist in characteristic types, each with a limited range of functional plasticity. Motor nerves or stimulation can only modify the phenotype of muscle fibers within that range. The particular range of the fiber is thus an intrinsic property determined genetically, it depends on the type of muscle and the genetic history of the fiber itself. This is consistent with the cross-innervation studies of Gordon and co-workers.[42] They cross-reinnervated a flexor and an extensor muscle group, so that the triceps surae muscle received input from the common peroneal nerve. The reinnervated soleus muscle produced no FG muscle fibers, although such fibers were found in the MG and lateral gastrocnemius muscles. A similar suggestion has been made by Gunning and Hardeman.[43] They suggest that embryonic cells destined to become muscle fibers differentiate under genetic control, a control that determines what types of myosin they will make. Slow and fast fibers differentiate early and become difficult to interconvert later. It is only at a later stage that the fast fiber types differentiate further. Environmental and functional information can influence the genetic control, however, so that adaptation is possible.

Bouchard and co-workers[44] studied the fiber composition in biopsy samples of muscles obtained from dizygotic and monozygotic twin brothers. The finding that identical twins showed greater similarity in their fiber type composition than did the other groups suggests a genetic component to muscle fiber composition. Considerable variability, however, remains. The system is flexible, and, although genetics clearly plays a role, it is not the whole story.

In animals and humans, the fiber composition of a muscle changes if its use is changed or if it is chronically stimulated. This was first shown by Salmons and Vrbova,[45] who were able to convert a muscle with a predominance of FG muscle fibers into a muscle composed only of SO fibers. Recently, this knowledge has been put into practice. Attempts have been made to convert muscles to serve a desired function quite deliberately. For example, the latissimus dorsi muscle, a muscle that in humans is rather fatigable, has been used as a pump to assist the heart, the ultimate fatigue-resistant muscle. In order to succeed in this, considerable information on the properties of muscles and their response to chronic stimulation had to be available (reviewed by Nemeth[39] and Kernell[46]). Paradoxically, although muscles may readily be rendered SO by chronic stimulation, the pattern needed to convert a muscle to one with a predominance of FG fibers is still not known, and studies are in progress to find one.[46] It may be that genetic determination plays a stronger role than has been suspected and that SO fibers are too stable to be converted completely.

Activity and Adaptation in Motor Units

Since the classic work of Buller et al,[47] it has been known that the speed, strength, and fatigue resistance of muscles and their motor units are not fixed properties; rather, they depend on the activity pattern the muscle undergoes. Buller and co-workers cut the nerves to the MG and soleus muscles and switched them, so that the MG muscle was reinnervated with the soleus muscle nerve, and vice versa. The muscles changed their properties in the direction determined by their new innervation: The soleus muscle became faster, and the MG muscle became slower. Buller and co-workers did not formulate the hypothesis explicitly in terms of activity. They also considered the possibility that a trophic influence of the nerve, rather than activity alone, might influence muscle fiber type. It remained for Salmons and Vrbova,[45] and later Lomo and co-workers,[48] to show that chronic stimulation of a muscle nerve, or of the muscle itself, alone would suffice to cause it to adopt properties of speed and fatigue resistance similar to those of the soleus muscle. As we have seen, however, the attempts to produce a conversion from slow to fast muscle fiber types have met with far less success. Kernell and co-workers were unable to show a clear correlation between activity patterns and motor unit strength and speed in chronically stimulated motor units; it proved to be easy to slow a muscle down and weaken it, whereas it was difficult or impossible to speed up or strengthen the muscle. Their extensive efforts have recently been reviewed by Kernell.[46]

Activity is likely a factor that determines the strength, fatigability, and speed of motor units. Thus, the observation that motor units are recruited in order of increasing force production must be interpreted with care, because it is possible that the recruitment order comes first and force production adapts to it.

The fiber type composition of the appropriate muscles in trained athletes appears to be adjusted to the tasks they perform (see Tab. 9 in the chapter by Saltin and Gollnick[14]). Thus, the lower-extremity muscles of sprinters are composed predominantly of fast-twitch fibers, whereas slow-twitch fibers predominate in long-distance runners. Similarly, the deltoid muscles of elite canoeists contain a predominance of slow-twitch fibers. It may be argued that these differences are genetic, and that genetically endowed persons become athletes and choose the appropriate sport. Although there is a significant element of truth in this possibility, plasticity furthered by training can also play a major role.

In humans, training can produce changes in muscle strength, speed, endurance, and fiber type composition. It is generally easier to change the metabolic properties of a muscle than its contractile properties.[16,19] The large body of literature on the effect of training has been reviewed. Endurance training appears to readily produce changes in the oxidative capacity of muscles. Thus, for example, the oxidative capacity of type I fibers may increase; in this respect, these fibers become more like IIa fibers. The contractile speeds need not change, however, and the distinction between I and IIa fibers may remain clear. This adds to the difficulty, mentioned earlier, of reconciling the fiber type classifications SO, FOG, and FG and I, IIa, and IIb. Not only does this difficulty in reconciling fiber type classifications create difficulties between animals and humans, it can create ambiguities in distinguishing fiber types between well-trained and sedentary humans. Fiber type conversions, such as those that occur during chronic stimulation, have been shown in longitudinal studies on the same person.[16]

The training required to produce changes in fiber types may need to be more rigorous and enduring than that usually used if fiber changes from IIb to IIa and from IIa to I are to be produced to a large degree. Even training of several hours a day is a far lesser demand than chronic stimulation, which goes on around the clock.

Changes in fiber size and type can be induced by certain hormones. Important among these are the thyroid hormones and testosterone.[19] This is the basis of doping to improve athletic performance.

Motor unit strength is not determined solely by the strength of its muscle fibers, but also by the number of muscle fibers the nerve fiber innervates. It is unlikely that this factor changes with activity, although it can change with reinnervation.[46] The "innervation ratio" has been calculated in a number of motor units. Usually, this ratio is determined by the percentages of the different muscle fiber types[49] in the muscle based on histological data, the total number of motoneurons innervating the muscle, and the percentages of each motor unit type.[49] This indirect method showed Burke[49] to calculate relative innervation ratios for a variety of muscles. The most extreme values given are for the TA muscle, in which FF motor units have 2.74 times as many fibers per unit than do S motor units. In a number of other muscles, the difference between innervation ratios of FF and S units is less than 2 to 1. These muscles include the MG (1.2:1), flexor digitorum longus (1.83:1), tibialis posterior (1.28:1), TA (1.33:1), and peroneus longus (1.21:1). If all muscle fibers had the same strength, this should also be the force difference between the FF and S motor units. We have seen that this is not so.

A third factor determining the strength of a motor unit is the "specific tension" of its muscle fibers, the force produced per cross-sectional area by the fiber (for a discussion of this issue, see Burke[49]). This is a controversial subject. If specific tension is assumed to differ among motor unit types, it can be argued whether specific tension changes with activity or with reinnervation of a muscle. Burke, using indirect methods, calculated that the specific force differed systemically with muscle fiber type. An experiment to measure specific tension directly suggested that there is no difference among muscle fiber types,[50] which is supported by indirect measurements obtained by Stein et al.[51] This issue is by no means settled, and may provide important information on the causes of force differences among the diverse motor unit types seen normally, after reinnervation, or when they have been chronically stimulated.

It is interesting to note the extent of recovery of a reinnervated muscle and the degree to which the orderly recruitment of motor units reestablishes itself. This occurs when a muscle is denervated and reinnervated with the same muscle nerve. Initially, the normal relationships between motoneuron size and motor unit strength and speed are absent, but they return with time.[52] Because nerve fibers do not reinnervate their original muscle fibers, a respecification must take place, as also occurs with chronic stimulation.

Such a respecification occurs even in the extreme case in which an extensor muscle group is reinnervated with a nerve normally supplying a flexor muscle.[42] In this case, muscle force returns to normal for that muscle, exceeding that of the flexor muscles by a factor of 2 to 3. Speed is comparable to that of the flexor muscles, consistent with what has been discussed previously. The nature of the variable that is respecified, however, is controversial. As discussed previously, motor unit strength depends on the number of muscle fibers innervated, their cross-sectional areas, and the force per unit area (specific tension) each fiber produces. The results of Gordon et al[42] Suggested that the specific tension was the major variable that changed. Yet, it is not at a clear that specific tension even differs with fiber type, as the authors point out.[42] In fact, opposite views on this issue are presented by Stein et al[51] and by Burke.[49]

The adaptability of muscle and its motor units has great relevance in physical therapy. This adaptability should form the basis for therapeutic measures to restore muscle strength after injury to the muscle or its nerve; inactivity produced by injury, casts, or splints; or central nervous system disease or lesions such as stroke. The relevance may be extended to functional electrical stimulation (FES). If a nerve is stimulated by implanted or percutaneous electrodes, increased stimulus strength is likely to recruit motor units roughly in the reverse of the normal recruitment order.[53] If the electrode lies within the muscle, the situation is harder to define. It is likely that no consistent recruitment order will be found (Stuart Binder-Macleod, PhD, PT, and colleagues; unpublished research). If FES is applied chronically, however, we may speculate that the stimulated muscle units may adapt to their imposed activity patterns and change their speed and strength characteristics. What effect this may have on the smoothness of induced movements is not known; at present, FES is in such an early stage of development that the question is not yet of great importance. To my knowledge, the question "What are the consequences of a disorderly recruitment order on the ability to grade muscle force smoothly?" has not received critical attention. When the issue is discussed, it is addressed as I have done, by showing that the cumulative force curve becomes less smooth. How important this is for standing or walking is not known.

Rate Coding

The force of a motor unit, or of a whole muscle, ranges from a minimum (ie, the twitch force) to a maximum (ie, the force produced by tetanic stimulation). This range may be expressed as tetanic tension divided by twitch tension, the tetanus:twitch ratio. The reciprocal of this value, the twitch:tetanus ratio, is frequently given in the literature. Figure 3 shows the twitch force plotted against tetanic force for motor units of each type for a variety of different muscles in the cat. These muscles include the diaphragm, MG, TA, EDL, personeus lorigus, abductor cruris caudalis, and rectus lateralis (a muscle that moves the eye). The slope of a straight line through these points gives the twitch: tetanus ratio. Figure 3 shows a roughly linear relationship between twitch and tetanic tension, and the actual values show no consistent difference between motor unit types in this regard. In part, this may be due to the fact that twitch tension is a notoriously variable quantity, because it changes rapidly with repetitive stimulation, showing potentiation and, in FF units, the beginnings of fatigue. Studies on potentiated motor units show a lower tetanus:twitch ratio than those on unpotentiated units, and it is not always stated how the studies were done.

The tetanus:twitch ratio is generally said to range from about 2 to 15; the data from which Figure 3 was made suggest that 5 is a reasonable mean value. Thus, rate coding, or varying the frequency at which motor units are driven, allows force to be varied by a factor of 5. That is not very much. Clearly, the additional feature of recruitment of more motor units is needed to provide the full range of tensions needed for normal movements.

Rate coding has the advantage over recruitment that the force gradation is smooth. Recruitment, on the other hand, must occur in steps. The two mechanisms likely occur together. Although it was suggested many years ago that rate coding is the predominant mechanism for varying muscle force,[54,55] the situation is not quite so simple. Kukulka and Clamann[56] were able to show that recruitment of new motor units occurred throughout the range of forces produced by the brachial biceps muscle in humans. In contrast, they were unable to show the recruitment of more motor units in the adductor pollicis muscle after that muscle produced about 50% of its maximum force; the remainder of the force had to be produced by rate coding, that is, by increasing the frequency with which the already active motor units were driven. This finding was in agreement with the earlier work of Milner-Brown et al,[54] who had also studied a muscle of the hand. The relative importance of rate coding and recruitment probably depends on the muscle and its function, although this problem has not been studied systematically.

Recruitment of Different Motor Units

The greatest range of forces may be produced by recruiting more motor units. The weakest motor unit of a typical skeletal muscle such as the cat MG muscle may produce a force of about 0.5 g,[5] whereas the whole muscle produces a maximum tetanic force of almost 13 kg.[10] This is a range of more than 20,000 to 1, which we can compare with that possible with rate coding, between 2 and 15 to 1. Zajac and Faden[57] have shown that MG muscle motor units are recruited in order of increasing force, from the least to the most powerful. Although the mechanism that determines the recruitment order is not yet understood, this finding is supported by other studies.[10,58] This orderly recruitment pattern has interesting consequences, for it leads to a convenient means of regulating muscle force referred to as "proportional control." In proportional control, a variable such as force is adjusted in steps that are proportional to the force already present. A weak force is graded by changing it slightly (eg, by steps of 5%), and a stronger force is graded using larger absolute increments, but still by 5% of the force already present. This means of regulating muscle force is accomplished by recruiting motor units in order of increasing strength.

The benefit of recruiting more motor units from weak to strong was already recognized by Henneman et al.[59] When total force output is weak, as it is when few motor units are active, force is increased by the recruitment of additional weak motor units. The steps in force produced by this recruitment pattern are small, and force gradation is rather smooth, As force increases, the motor units remaining to be recruited are progressively more powerful. In an ideal proportional control system, the force added by the recruitment of each new motor unit is a fixed percentage of the force already present. if we plot force output against the number of active motor units, we obtain an exponential curve. The force increase of muscles resembles such a curve rather closely, but not perfectly. As Henneman et al pointed out,

... this grading of output is reminiscent

of the Weber fractions known in

the field of sensory discrimination. We

are not proposing a precise mathematical

relationship for the motor system

comparable to Weber's rule of the just

noticeable difference of sensation, but

the analogy is clear: the smallest increment

that can be added to the force

exerted by a muscle becomes the

greater as the force of contraction

increases.[59(p578)] A brief digression may be appropriate here. Further information on sensory physiology may be found in physiology texts.[1,3,4] Weber was a sensory physiologist of the last century who was interested in the accuracy of our ability to "measure" sensations. His first experiments involved the ability of a subject to judge weights. The same principle, however, applies to the ability to judge the brightness of light, the intensity of sound, the pressure on the skin, and, as discussed earlier, the ability to adjust force output. Weber found that the ability to sense differences in a stimulus depended on the strength of the stimulus. Thus, it was as easy to distinguish a 1-g change in a 10-g weight as it was to notice a 100-g change in a 1-kg weight, a 10% change in each instance. The smallest such detectable change bust noticeable difference) could be expressed as a fraction, the quotient of the change in stimulus strength ([delta]s) divided by the stimulus strength (s). This proved to be a constant (K) over a wide range of stimulus strengths, leading to Weber's law: [delta]s/s=k. In this example, if the just noticeable difference is actually 10%, a 1-g change, or even a 10-g change, in a kilogram weight could not be detected, although it is readily detected in, a 10-g weight.

The perceived magnitude of a change in the intensity of a stimulus (eg, loudness of a sound, heaviness of a weight) depends on the size of the stimulus that is changed. The same principle applies for changes in force output. By recruiting weak motor units to increase a weak force, roughly the same percentage of change is produced as when powerful motor units are recruited to increase a powerful force. This way of grading force is practical: The recruitment of one or two weak quadriceps femoris muscle motor units producing 50 mN (5g) of force will hardly affect the accuracy (or anything else) of a basketball player's slam dunk. Large forces are graded by the recruitment of large motor units.

Motor units display a range of forces, even in a muscle as homogeneous as the soleus muscle (31-1,630 mN, or 3.2-40.4 g, of force[37]). This may be incorporated into a model of recruitment by assuming that the forces of the population of motor units are distributed according to Gaussian statistics. The distribution within each motor unit type is probably skewed, with a preponderance of weaker units, but, to my knowledge, data are not available. If the motor units of such a population are recruited at random, force win increase in a more or less straight line as a function of the number of motor units recruited.[58] Individual increments will be large or small, so that the line will have a lot of jumps or kinks, but, on average, it will resemble a straight line rather than a smooth curve. Figure 4 presents force against the number of active motor units for a model soleus muscle. The maximum force has been normalized to 1, and the number of motor units has been normalized to 100. The upper curve shows the results of a random recruitment order.

If we rank order the population of motor units according to increasing force and recruit them in that rank order, the force rises in a manner more closely predicted by Weber's law (Fig. 4, curved line). A system in which force is graded in proportion to the force already present is said to be under proportional control. The recruitment of motor units by their size and their progressively increasing strengths ensures proportional control.

A far more interesting example is that of a mixed muscle. I have chosen to model two such muscles, the EDL and the FCR. These muscles were chosen because of their extremely different distributions of motor unit types: The EDL muscle has 6% type S motor units and 48% type FF motor units, whereas the FCR muscle has 40% type S motor units and only 22% type FF motor units.

Figure 5 shows the modeled recruitment curves for these two muscles superimposed, as linear plots in the upper panel and as semilogarithmic plots in the lower panel. The linear plots are smooth in shape, rising slowly at first, then more and more steeply, as would be expected from a proportional control system. In addition, because the model allows considerable scatter in motor unit force, as does an actual muscle, motor units are not recruited according to type when they are recruited in order of increasing strengths.[57] Some type S motor units are stronger than some type FR motor units and are assumed to be recruited later. Thus, the transition from one type of motor unit to the next is smoothed. we have the remarkable result that scatter in the force output in the population of each motor unit type (ie, the randomness of a Gaussian distribution) produces a smoothing of force gradation.

We can rescale the ordinate to a logarithmic scale, converting the exponential rise into a roughly linear rise. The graph becomes approximately linear for high force levels, but not for low forces. This is consistent with recruitment curves for a variety of real muscles, such as MG,[58] FCR,[22] and human thenar muscles.[34,35]

A tangent to the curve on a linear scale gives the force change produced per motor unit recruited. As the curve becomes progressively steeper, the recruitment of a few additional units produces a large change in force. Figure 5 (top) shows that about 70% of the motor units are required to produce half of the EDL muscle's maximum force, whereas about 75% of the motor units are needed to produce half of the FCR muscle's force., Conversely, half of the motor units of the FCR muscle, with its large number of S motor units, produce only 10% of its maximum force, whereas half of the EDL muscle's motor units produce about 25% of its force.

A number of sophisticated models of motor unit recruitment have been proposed in which additional factors such as rate coding for force modulation, the properties of reflex inputs, and motoneuron properties are also considered. The most detailed recruitment models are those of Kernell[46] and Heckman and Binder.[60] Heckman and Binder used tetanic force distributions for whole muscles obtained from the literature without considering muscle unit types explicitly. This was convenient, because their model is of the cat MG muscle only. A major addition to such models was the incorporation of rate coding. They thus created a muscle model rather like the biceps brachii muscle model described by Kukulka and Clamann,[56] in which recruitment and rate coding occur throughout the muscle force range. The intent of this model is to create an input-output relationship for a muscle, in which the input is the afferent drive to the motoneuron pool and the output is the force. In principle, this model could be used to examine the relationship for a reinnervated muscle and to determine how smoothly force is graded.

Solely to obtain proportional force gradation, it appears to be desirable for a muscle to possess a population of motor units with a range of forces. Such an arrangement seems to be superior to that of a muscle consisting of a population of identical motor units. But why is it desirable to have motor units of discretely different capabilities (ie, motor units of different types)?

As Henneman and Olson[8] pointed out long ago, one could expect muscles serving very different functions to develop different capabilities. They then raised the more difficult issue of two closely synergistic muscles, the gastrocnemius and the soleus, which perform the same function to extend the foot, and even insert onto the same tendon. Long muscle fibers forming relatively weak, fatigue-resistant motor units make the soleus muscle ideally designed to make slow, finely graded movements and to shorten over a considerable length. The gastrocnemius muscle, with its pennate fiber arrangement and powerful, fatigable motor units, is designed to do just the opposite: provide powerful, fast contractions with less endurance, and over a shorter range of lengths. Henneman and Olson concluded that nature decided that to move the ankle, two heads are better than one.

The problem becomes more difficult in a single mixed muscle, because the architecture of the individual motor units is likely to be the same. Thus, all motor units will be involved in a pennate, or perhaps fusiform, construction, and all motor units will be forced to make contractions over the same length and will usually work at the same mechanical advantage (a well-known exception is the cat sartorius muscle[61]). Even in such a homogeneous environment, however, it is difficult to envision a compromise motor unit type that can perform all functions reasonably well. Although FR motor units may appear to offer such a compromise, the Table shows that they are less than half as strong as FF units. This loss of strength could be dearly bought in trying to flee from a predator. It is well known that sprinters and marathon runners train their motor units differently and are unwilling to produce compromise muscles by training.

Clearly, strength requires muscle fibers fined with contractile structures, leaving little room for the storage of energy-rich molecules and mitochondria, the chemical factories that make the energy available. Type FF muscle fibers are designed this way, and their large diameter and poor blood supply make them well suited for powerful contractions that fatigue rapidly. When a muscle produces a powerful contraction, it generates considerable internal pressure and cuts off its own blood supply.[46,62] Hence, powerful muscle fibers cannot be resupplied with oxygen and nutrients and are dependent on their own stores and anaerobic metabolism. Type FF motor units are ideally suited for this function.

Oxidative motor units are weaker and their muscle fibers are of smaller diameter than are anaerobic (FG) muscle fibers in most species (they are more homogeneous in humans, but still tend to be smaller[14]; however, see Simoneau and Bouchard[40]); much of their intracellular space is devoted to the production of energy. These motor units are also richly supplied with capillaries. Because these motor units are unlikely to cut off their own blood supply with powerful contractions, they can receive oxygen and nutrients and function for long periods of time. The fibers of these motor units are scattered throughout a cross-section of the muscle, and not arranged in clumps. This construction prevents the generation of local muscle pressure, which would tend to block blood flow. Hence, the interdigitation of motor units in a normal muscle and their asynchronous discharge help support the supply of oxygen and nutrients. It is difficult to conceive of a compromise motor unit that would perform the different functions of powerful contractions on the one hand and steady, sustained contractions on the other equally well or even adequately.

Most muscle exertions, especially very strong ones, tend to be phasic; running and even walking are good examples For such rhythmic force output, blood flow is not occluded, or occluded only briefly, and may even be assisted, as venous return is facilitated by muscle activity. An additional consideration is that the activities we can perform for long periods of time, such as walking, are probably performed by fatigue-resistant motor units alone. Thus, for example, the walking cat recruits 25% of its MG muscle and even less of its lateral gastrocnemius muscle when walking[63] (Table), well within the range of fatigue-resistant motor units.

It is clear that the knowledge of the properties of muscles and their motor units is applied in athletics and therapeutics, and forms a core of the knowledge influencing such new techniques as FES. As more knowledge of muscle function and muscle organization emerges from basic science and clinical laboratories, it will have a profound influence on the nature of the therapeutic process.

References

[1] Guyton AC. Textbook of Medical Physiology. 7th ed. Philadelphia, Pa: WB Saunders Co; 1986: chap 13. [2] Berne RM, Levy MN. Electrical activity of the heart. In: Berne RM, Levy MN, eds. Physiology. 2nd ed. St Louis, Mo: CV Mosby Co; 1988: chap 27. [3] Milnor WR. Properties of cardiac tissue. In: Mountcastle VB, ed. Medical Physiology. 14th ed. St Louis, Mo: CV Mosby Co; 1980: chap 36. [4] Antoni H; Biederman-Thorson M, trans. Function of the heart. In: Schmidt RF, Thews G, eds. Human Physiology. 2nd ed. New York, NY: Springer-Verlag New York Inc; 1989: chap 19. [5] Wuerker RB, McPhedran AM, Henneman E. Properties of motor units in a heterogeneous pale muscle (m. gastrocnemius) of the cat. J. Neurophysiol. 1965;28:85-99. [6] McPhedran AM, Wuerker RB, Henneman E. Properties of motor units in a homogeneous red muscle (soleus) of the cat. J Neurophysiol. 1965;28:71-84. [7] Henneman E, Mendell LM. Functional organization of motoneuron pool and its inputs. in: Brooks VB, ed. Handbook of Physiology, Volume 2. Motor Control. Bethesda, Md: American Physiological Society; 1981: chap 11. [8] Henneman E, Olson CB. Relations between structure and function in the design of skeletal muscles. J Neurophysiol, 1965;28:581-598. [9] Edstrom L, Kugelberg E. Histochemical composition, distribution of fibers and fatigability of single motor units: anterior tibial muscle of the rat. J Neurol Neurosurg Psychiatry. 1968; 31:424-433. [10] Burke RE. Motor units: anatomy, physiology, and functional organization. In: Brooks VB, ed. Handbook of Physiology, Volume 2: Motor Control Bethesda, Md: American Physiological Society; 1981: chap 10. [11] McDonagh JC, Binder MD, Reinking RM, Stuart DG. Tetrapartite classification of motor units of cat tibialis posterior. J Neurophysiol. 1980;44:696-712. [12] Brooke MH, Kaiser KK. Three "myosin adenosine triphosphatase" systems: the nature of their pH lability and sulfhydryl dependence. J Histochem Cytochem. 1970;18:670-672. [13] Garnett RAF, O'Donovan MJ, Stephens JA, Taylor A. Motor unit organization of human medial gastrocnemius. J Physiol (Lond). 1978; 287:33-43. [14] Saltin B, Gollnick PD. Skeletal muscle adaptability: significance for metabolism and performance. In: Peachey LD, ed. Handbook of Physiology. Bethesda, Md: American Physiological Society; 1983: chap 19. [15] Pette D, Vrbova G. Neural control of phenotypic expression in mammalian muscle fibers. Muscle Nerve 1985;8:676-689. [16] Howald H, Hoppeler H, Claassen H, et al. influences of endurance training on the ultrastructural composition of the different muscle fiber types in humans. Pflugers Arch. 1985;403: 369-376. [17] Brooke MH, Kaiser KK. The use and abuse of muscle histochemistry. Ann NY Acad Sci. 1974;228:121-144. [18] Peter JB, Barnard RJ, Edgerton VR, et al. Metabolic profiles of the three fiber types of skeletal muscle in guinea pigs and rabbits. Biochemistry. 1972;11:2627-2633. [19] Billeter R, Oetliker H, Hoppeler H. Structural basis of muscle performance. In: Jones JH, ed. Comparative Vertebrate Exercise Physiology, Volume 37 Advances in Veterinary Science and Comparative Medicine. San Diego, Calif.. Academic Press Inc. In press. [20] Romanul FCA. Enzymes in muscle, I: histochemical studies of enzymes in individual muscle fibers. Arch Neurol. 1964;11:355-368. [21] Lowry CV, Kimmey JS, Felder S, et al. Enzyme patterns in single human muscle fibers. J Biol Chem. 1978;253:8269-8277. [22] Botterman BR, Iwamoto GA, Gonyea WJ. Classification of motor units in flexor carpi radialis. J Neurophysiol. 1985;54:676-690, [23] Dum RP, Kennedy IT. Physiological and histochemical characteristics of motor units in cat tibialis anterior and extensor digitorum longus muscles. J Neurophysiol. 1980;43:1615-1630. [24] Ariano MA, Armstrong RB, Edgerton VR. Hindlimb muscle fiber populations of five mammals J Histiochem Cytochem. 1973;21:51-55. [25] Davies AS, Gunn HM. Histochemical fibre types in the mammalian diaphragm. J Anat. 1972;112:41-60. [26] Johnson MA, Polgar J, Weightman D, Appleton D. Data on the distribution of fibre types in thirty-six human muscles: an autopsy study. J Neurol Sci. 1973;18:111-129. [27] Goldberg SJ. Mechanical properties of extraocular motor units. In: Binder MD, Mendell LM, eds. The Segmental Motor System. New York, NY: Oxford University Press Inc; 1990: chap 12. [28] Shall MS, Goldberg SJ. Extraocular motor units: type classification and motoneuron stimulation frequency-muscle unit force relationships. Brain Res. 1992;587:291-300. [29] Kernell D, Eerbeek O, Verhey BA. Motor unit categorization on the basis of contractile properties: an experimental analysis of the composition of the cat's m. peroneus longus. Exp Brain Res. 1983;50:211-219. [30] Buchthal F, Schmalbruch H. Contraction times and fibre types in intact human muscle. Acta Physiol Scand. 1970;79:435-452. [31] Sica REP, McComas AJ. Fast and slow twitch units in a human muscle. J Neurol Neurosurg Psychiatry. 1971;34:113-120. [32] Taylor A, Stephens JA. Study of human motor unit contractions by controlled intramuscular microstimulation. Brain Res. 1976;117:331-335. [33] Milner-Brown HS, Stein RB, Yemm R. The contractile properties of human motor units during isometric voluntary contractions. J Physiol (Lond). 1973;228:285-306. [34] Thomas CK, Bigland-Ritchie B, Westling G, Johansson RS. A comparison of human thenar motor-unit properties studied by intraneural motor-axon stimulation and spike-triggered averaging. J Neurophysiol. 1990;64:1347-1351. [35] Westling G, Johansson RS, Thomas CK, Bigland-Ritchie B. Measurement of contractile and electrical properties of single human thenar motor units in response to intraneural motor-axon stimulation. J Neurophysiol. 1990; 64:1331-1338. [36] Thomas CK, Johansson RS, Westling G, Bigland-Ritchie B. Twitch properties of human thenar motor units measured in response to intraneural motor-axon stimulation. J Neurophysiol. 1990;64:1339-1346. [37] McDonagh JC, Binder MD, Reinking RM, Stuart DG. A commentary on muscle unit properties in cat hindlimb muscles. J Morphol. 1980;166:217-230. [38] George JC, Berger AJ. Avian Myology. New York, NY.. Academic Press Inc; 1966: chap 4. [39] Nemeth PM. Metabolic fiber types and influences on their transformation. In: Binder MD, Mendell LM, eds. The Segmental Motor System. New York, NY. Oxford University Press; 1990: chap 14. [40] Simoneau JA, Bouchard C. Human variation in skeletal muscle fiber-type proportion and enzyme activities. Am J Physiol. 1989;257: E567-E572. [41] Hoh JFY. Myogenic regulation of mammalian skeletal muscle fibres. News in Physiological Sciences. 1991;6:1-6. [42] Gordon T, Thomas CK, Stein RB, Erdebil S. Comparison of physiological and histochemical properties of motor units after cross-reinnervation of antagonistic muscles in the cat hindlimb. J Neurophysiol. 1988;60:365-378. [43] Gunning P, Hardeman E. Multiple mechanisms regulate muscle fiber diversity. FASEB J. 1991;5:3064-3070. [44] Bouchard C, Simoneau JA, Lortie G.. Genetic effects in human skeletal muscle fiber type distribution and enzyme activities. Can J Physiol Pharmacol. 1986;64:1245-1251. [45] Salmons S, Vrbova G. The influence of activity on some contractile characteristics of mammalian fast and slow muscles. J Physiol (Lond). 1969;201:535-549. [46] Kernell D. Organized variability in the neuromuscular system: a survey of task-related adaptations. Arch Ital Biol. 1992;130:19-66. [47] Buller AJ, Eccles JC, Eccles RM. Interactions between motoneurones and muscles in respect of the characteristic speeds of their responses. J Physiol (Lond). 1960;150:417-439. [48] Lomo T, Westgaard RH, Dahl HA. Contractile properties of muscle: control by pattern of muscle activity in the rat. Proc R Soc Lond [Biol]. 1974;187:99-103. [49] Burke RE. Motor unit types: some history and unsettled issues. In: Binder MD, Mendell LM, eds. The Segmental Motor System. New York, NY. Oxford University Press Inc; 1990: chap 11. [50] Lucas SM, Ruff RL, Binder MD. Specific tension measurements in single soleus and medial gastrocnemius muscle fibers of the cat. Exp Neurol. 1987;95:142-154. [51] Stein RB, Gordon T, Totosy de Zepetnek J. mechanisms for respecifying muscle properties following reinnervation. In: Binder MD, Mendell LM, eds. The Segmental Motor System. New York, NY: Oxford University Press inc; 1990: chap 15. [52] Gordon T, Stein RB. Reorganization of motor-unit properties in reinnervated muscles of the cat. J Neurophysiol, 1982;48:1175-1190. [53] Clamann HP, Gillies JD, Skinner RD, Henneman E. Quantitative measures of output of a motoneuron pool during monosynaptic reflexes. J Neurophysiol. 1974;37:1328-1337. [54] Milner-Brown HS, Stein RB, Yemm R. The orderly recruitment of human motor units during voluntary isometric contractions. J Physiol (Lond). 1973;228:359-370. [55] Milner-Brown HS, Stein RB, Yemm I;L Changes in firing rate of human motor units during linearly changing voluntary contractions. J Physiol (Lond). 1973;228:371-390. [56] Kukulka CG, Clamann HP. Comparison of the recruitment and discharge properties of motor units in human brachial biceps and adductor pollicis during isometric contractions. Brain Res. 1981;219:45-55. [57] Zajac FE, Faden JS. Relationship among recruitment order, axonal conduction velocity, and muscle-unit properties of type-identified motor units in cat plantaris muscle. J Neurophysiol. 1985;53:1303-1322. [58] Fleshman JW, Munson JB, Sypert GW, Friedman WA. Rheobase, input resistance, and motor-unit type in medial gastrocnemius motoneurons in the cat. J Neurophysiol. 1981;46: 1326-1338. [59] Henneman E, Somjen G, Carpenter DO. Functional significance of cell size in spinal motoneurons. J Neurophysiol. 1965;28:560-580. [60] Heckman CJ, Binder MD. Computer simulation of the steady-state input-output function of the cat medial gastrocnemius motoneuron pool. J Neurophysiol 1991;65:952-967. [61] Hoffer JA, Loeb GE, Sugano N, et al. Cat hindlimb motoneurons during locomotion, III: functional segregation in sartorius. J Neurophysiol. 1987;57:554-562. [62] Salmons S. Functional adaptation in skeletal muscle. In: Evarts EV, Wise SP, Bousfield D, eds. The Motor System in Neurobiology. Amsterdam, the Netherlands: Elsevier Science Publishers BV; 1985:23-29. [63] Loeb GE, He J, Levine WS. Spinal cord circuits: Are they mirrors of musculoskeletal mechanics? Journal of Motor Bevavior. 1989;21: 473-491. [64] Fournier M, Sieck C. Mechanical properties of muscle units in the cat diaphragm. J Neurophysiol. 1988;59:1055-1066.

HP Clamann, PhD, is Professor of Anatomy, Department of Physiology, University of Berne, Buhlplatz 5, CH 3012 Berne, Switzerland.

This work was supported in part by a grant from the Swiss National Science Foundation.