HUMAN EISEVIER
Human
Architectural
Movement
Science
13 (1994) 545-556
features of multiarticular
F!!i!!NT muscles
Gerald E. Loeb *, Frances J.R. Richmond MRC Group in Sensory-Motor Physiology and Bio-Medical Engineering Unit, Queen’s University, Kingston, Ontario, Canada K7L 3N6
Abstract
Most of the available information on neuromuscular organization and musculoskeletal mechanics comes from a few muscles that physiologists choose to study because they are relatively simple. They usually cross a single hinge joint, have a single band of innervation, and are composed of muscle fibres with similar lengths, all acting in parallel. In most species, such muscles are in the minority. In order to develop quantitative models of complete musculoskeletal systems, it is necessary also to study the less simple muscles. This has revealed several types of architecture that were somewhat unexpected and may have important functional consequences. Limb muscles that span more than one joint are often quite long and must withstand large length changes; their parallel fascicles often contain multiple, relatively short fibres interdigitated in-series. Axial muscles often span many joints and are composed of in-series compartments. A variety of neuromuscular architectures seems to have evolved to deal with problems of mechanical instability arising from imbalances of force between fibres, motor units and compartments. These may be prone to particular failure modes in response to athletic strain, dystrophic diseases, and functional electrical stimulation.
1. Introduction
In neurophysiological textbooks, individual muscles are commonly assumed to be the primary units of organization in the planning and control
* Corresponding
author.
Tel.:
+ 1 613 545-2790;
0167-9457/94/$07.00 0 1994 Elsevier SSDZ 0167-9457(94)00029-8
Science
Fax: + 1 613 545-6802.
B.V. All rights reserved
546
G. E. Loeb, F.J.R. Richmond /Human
Movement Science 13 (19941545-556
of movement by the central nervous system. This idea seems consistent with features found at several different levels of structure. Using gross anatomical dissection, the musculature of most parts of the body is readily divided into a number of fairly discrete entities separated by fascial planes. The fascicles within each such entity often appear to have similar lengths and orientations (angle of pennation). Each of these entities is supplied by its own muscle nerve or nerves, defining neuromuscular compartments (see Chanaud et al., 1991a,b, for review and definitions). The cell bodies of motoneurons serving each neuromuscular compartment are usually clustered into discrete, columnar motor nuclei in the ventral horn of the spinal cord. The motoneurons in each motor nucleus tend to receive similar sets of descending and sensory input, causing them to be recruited together in an orderly manner depending on the amount of force required to perform a task. These facts have often been interpreted to suggest that relatively few morphometric parameters are needed to describe adequately the architecture of any muscle. The few muscles that have been studied in detail as model systems for spinal neurophysiology were all selected because they have particularly simple architectures that facilitate the design and interpretation of experiments. For example, the muscle fibres of the soleus muscle of the cat hindlimb span from bone to aponeurosis; all of its fibres have a similar length and mechanical action in the principal axis of a single joint; a single muscle-nerve supplies a single band of endplates. In such a simple muscle, it is perhaps not surprising that the afferent and efferent neurons and the reflex connections between them conform to the simple size and servocontrol principles for recruitment that are described in neurophysiological textbooks. In the past few years, a number of investigators have sought to understand the performance of complete, natural motor behaviours such as walking, bicycling and reaching. The first step is often to model the musculoskeletal system as a set of individual muscles. Each muscle is supposed to have a singular skeletal action (or set of actions if multiarticular), a single fibre-architecture that governs its ability to generate force under different conditions of length and velocity, and a unidimensional level of recruitment which can be measured by recording EMG at any single site in or over the muscle. Unfortunately, close inspection of the majority of muscles usually leads to confusion about how to proceed with either models or measurements. Many muscles have architectural features and internal heterogeneities that cannot be found in the idealized muscles
G.E. Loeb, F.J.R. Richmond /Human
Movement Science 13 (1994) 545-556
541
that form the basis of textbook models. Muscles that cross more than one joint seem to be the most difficult, for reasons that are probably related to functional requirements as well as to phylogenetic and ontogenetic development. These multiarticular muscles tend to be very long or very wide or to follow tortuous paths. The following is a brief description of some of the architectural features that tend to occur in such muscles, along with speculations about their possible mechanical function and implications for sensorimotor control and pathophysiology.
Muscles that cross more than one joint tend to be long, but this length can be achieved in different ways: (1) Distal muscles often achieve their spans via long tendons attached to relatively short muscle bellies. Their fibre architecture is often highly pennate, presumably to compensate for the relatively small moment arms (and consequently small mechanical advantage) provided by tendons that pass close to the centres of rotation of the joints (Young et al., 1993). The muscle mass is located proximally to minimize inertial mass in the distal portion of the limb, which often must be translated rapidly through space, as during locomotion. (2) Proximal limb muscles are usually designed quite differently in order to maximize torque and power output (Gans, 19821. Their attachments are often far from the centres-of-rotation of the joints that they cross, providing large moment arms and moments. This results in proportionately large length changes, particularly when the joints that they cross change angular position in ways that add constructively in terms of the length changes of the muscle. In order to avoid over-stretching their muscle fibres, such muscles must be nonpennate, with parallel muscle fascicles that run from origin to insertion, often with relatively little interposed tendon or aponeurotic sheet. (3) Multiarticular axial muscles appear to be derived from the fusion of primordial segmental muscles, often with vestigial features such as multiple, segmentally arranged nerves and motor nuclei and tendinous inscriptions between compartmentalized muscle fibres (Richmond et al., 1985; Richmond and Vidal, 1988). They have the most complex mechanical actions, often acting across five or six joints, each of which may have multiple degrees of freedom. Their moment arms at each joint appear to be complexly dependent on the position of multiple joints, although there are few quantitative studies (Selbie et al., 1993).
548
G.E. Loeb, F.J.R. Richmond /Human
2. Series-fibre architecture
Mouement Science I3 (1994) 545-556
for long spans
It has commonly been assumed that the individual fibres in muscles with long fascicles extend the length of those fascicles, particularly when there is no internal connective tissue such as tendinous inscriptions to provide an obvious terminus for shorter fibres. Exceptions to this assumption have been described intermittently for over a century; in fact, the assumption is usually wrong and the exceptions appear instead to be the rule (Loeb et al., 1987). In the cat, very few muscle fibres longer than 4 cm have been found, even in muscle fascicles that are 12-15 cm long; most are distributed around means of 2-3 cm in those long muscles that have been studied. Homologous muscles in goat (Gans et al., 1989) and giraffe (Gans and Loeb, unpublished observations) have fibres that are only modestly longer (perhaps 6-8 cm), so that a minimum of 6-20 fibres in-series is required to span fascicles up to a metre in length. Despite their modest size in the animal kingdom, humans have perhaps the longest muscle fibres yet to be described, up to 18 cm or more (Schwarzacher, 1959; Heron and Richmond, 1993). Even these long human fibres, however, fail to span the entire length of the fascicles in muscles such as sartorius or gracilis (Fig. 1). The intramuscular ends of the nonspanning fibres in both humans and subprimates taper gradually to a fine point; their sarcolemmal surfaces show no signs of the specialized mechanical attachments that are typical of myotendinous junctions (Trotter, 1990). Instead, they appear to dissipate the tensile forces produced by their contractions into the adjacent loose matrix of endomysial and perimysial connective tissue via shear forces that are distributed over the large surface area afforded by the long tapered end.
3. Mechanical stability in series-fibred muscles Why do long, parallel-fibred muscles have short, in-series fibres? Loeb et al. (1987) suggested that such an arrangement might be necessary to cope with the relatively slow conduction of action potentials along muscle fibres. During the rising phase of tension in a long muscle fibre, the central sarcomeres near the endplate would begin to generate tension and stretch the as-yet-passive distal sarcomeres. If these sacomeres are pulled beyond the peak of the length/tension relationship, the force output of the muscle would be ill-defined and unstable. If a motoneuron were to innervate a
G.E. Loeb, F.J.R. Richmond /Human
Movement Science 13 (1994) 545-556
549
Fig. 1. Fibre organization in the human leg muscle, sartorius: A: schematic location from origin on the ischium to insertion on the proximal tibia; B: schematic arrangement of fascicles spanning origin to insertion and consisting of relatively short, interdigitated fibres; C and D: photomicrographs of dissected single fibres. In C a tapered end of one fibre has been separated from the thicker shafts of neighbouring fibres; the boxed region is shown at a higher magnification in D, revealing cross-striations in both the thick and thin fibres. (Modified from Heron and Richmond, 1993.)
group of short fibres attached to each other in-series, they would all be activated simultaneously and quite homogeneously, producing an even distribution of tension along the fascicle. Unfortunately, there are at least three things wrong with this simple explanation. First, the tapered ends of fibres innervated by one motoneuron usually interdigitate among the fibres controlled by other motoneurons, so tension must pass at least some distance laterally through the connective
550
G.E. Loeb, F.J.R. Richmond /Human
Movement Science 13 (1994) 545-556
tissue and the intervening, often passive, muscle fibres (Pratt et al., 1986; Smits et al., 1991). Second, in at least one muscle that has been studied so far (feline anterior sartorius), many individual motoneurons innervate populations of muscle fibres that are distributed in highly asymmetrical patterns along the length of the muscle (Thomson et al., 1991; Smits et at., 1991). Finally, long muscles in humans appear to have an unusually broad distribution of fibre lengths, with many muscle fibres as long as 12-18 cm, yet the twitch dynamics of human muscles are only moderately slower than those of other animals with much shorter fibres and similar conduction velocities. One possible explanation for all of these phenomena requires the endomysial and perimysial connective tissue to play a much larger role in muscle mechanics than is usually assumed (Trotter, 1990). If local imbalances in tensile force could be dissipated laterally as shear forces anywhere along the length of any muscle fibre (active or passive), then the only requirement for maintaining regional homogeneity and stability of force output would be a grossly similar density of actively contracting fibres over the length of the fascicle. This could be achieved by spinal circuitry that tended to recruit similar numbers of motor units supplying each region of the muscle. The motoneurons innervating proximal and distal regions of anterior sartorius seem to be intermingled in their motor nuclei (Gordon et al., 1991). It seems unlikely that the balance of recruitment would always be perfect, but it is not clear how much imbalance would cause a problem. At low recruitment levels when active fibres are distributed sparsely, the collective passive tension of the inactive fibres would probably prevent significant shortening of active fibres in series with them. At higher recruitment, there would probably be at least some recruited fibres everywhere along the length of each fascicle. The regions with a lower density of active fibres would be able to produce less isometric force than other regions in series with them, but muscle has a very steep force-velocity relationship around zero (isometric) velocity (Joyce et al., 1969). Any tendency for some muscle fibres to contract at the expense of lengthening others would be opposed by an instantaneous rise in the tension of the stretched fibres (to about 1.6 times isometric) and a decrease in the shortening fibres (to about 0.8 times isometric). A similar mechanism has been proposed to operate when sarcomeres have heterogeneous lengths along single muscle fibres (Morgan, 1985). Unless the imbalance in recruitment were greater than two to one, little motion would actually occur before the end of a typical phasic contraction.
G.E. Loeb, F.J.R. Richmond/Human
Movement Science 13 (1994) 545-556
There are several situations in which this arrangement unstable, perhaps resulting in damage to the muscle:
551
might become
(1) The reliance on viscous and frictional coupling between fibres raises the possibility that they may sometimes slide to positions in which they exert poorly balanced shear forces on the endomysial and perimysial connective tissue. Hamstring pull injuries commonly occur during phasic (but not necessarily maximal) use of muscles that have not been adequately “warmed up” by passive stretching exercises (Burkett, 1970; Dornan, 1971; Liemohn, 1978). Their pathophysiology remains obscure but may be related to failure in shear between muscle fibres. (2) The ability of the sarcolemma to sustain these shear forces may be seriously compromised by muscular dystrophy, which appears to involve a defective connective protein that bridges the sarcolemma to the basal lamina (Ibraghimov-Beskrovnaya et al., 1992) and which commonly affects proximal limb muscles most severely (reviewed in Heron and Richmond, 1993). (3) This architecture may pose problems when muscles are stimulated not by their normal motoneuronal recruitment but through artificial means such as functional electrical stimulation (FES). FES is increasingly being applied to muscles that are paralyzed as a result of stroke and spinal cord injury, particularly to proximal limb muscles that are capable of producing the large moments required for locomotion. In muscles such as anterior sartorius, stimulation applied near only one of multiple nerve branches supplying asymmetrical territories can result in eccentric work in the sparsely innervated regions of the muscle (Scott et al., 1992). Eccentric exercise is known to cause damage in muscles that do not work this way normally, probably because of the increased tension that actively lengthening muscle fibres can produce. The consequences of such chronic stimulation should be examined before this new therapy is applied to muscles with in-series compartments in order to condition their strength or to reanimate paralyzed limbs.
4. Series-compartment
architecture
for multi-articular
spans
Not all long muscles that are composed of short muscle fibres have fascicles that run the entire length of the muscle. Some multiarticular muscles are composed of several discrete compartments of muscle fascicles
552
G.E. Loeb, F.J.R. Richmond/Human
Movement Science 13 (1994) 545-556
SPLENIUS
LONGISSIMUS CAPITI
Fig. 2. Anatomy and motor-unit organization of the neck muscle splenius: Top: line drawing to show its distributed origin along the nuchal midline raphe and insertion onto the lambdoidal crest of the skull; the lateral but not the medial part is divided by inscriptions into separate in-series fascicles. Bottom: schematic illustration of the distribution of muscle fibres innervated by muscle nerves from different cervical segments. Laterally, the tendinous inscriptions separate the motor units supplied by one segment from those supplied by adjacent segments; medially, motor units supplied by adjacent segments are also organized in series, but fibres composing the subpopulations of motor units interdigitate so that no distinct boundaries can be defined. (Modified from Richmond et al., 1985.)
that are attached at one or both ends to inscripted bands of connective tissue that cross much or all of the width of the muscle. The biarticular semitendinosus muscle of the hindlimb has two such discrete compartments in series, each with its own nerve composed of separate motoneurons that are largely intermingled in a single motor nucleus (English and Weeks, 1987). Long axial muscles such as those of the back and neck often span a large number of joints and may consists of a similar number of discrete, inscripted compartments, each innervated by a separate spinal nerve and innervated by motoneurons originating in segmentally circumscribed subnuclei of the overall motor nucleus (Gordon and Richmond, 1991). However, as shown in Fig. 2, the compartments defined by connective tissue and those defined by the muscle unit territories supplied by segmental
G.E. Loeb, F.J.R. Richmond /Human
Movement Science 13 (1994) 545-556
553
nerves are not necessarily coextensive (Loeb, 1989) and may consist of complex arrangements that are partially in-series and partially in-parallel. Many of the descending and segmental afferent pathways terminate quite unevenly in different spinal segments, raising the problem of maintaining balanced recruitment in the motor subnuclei supplying the various compartments of axial muscles. Nevertheless, balanced recruitment of the in-series compartments appears to be the rule (Richmond et al., 1992). Ensuring such a balanced force output would seem to be a particularly suitable application for servocontrol from the numerous (Richmond and Bakker, 1982) and highly elaborated (Richmond et al., 1986) proprioceptors of these muscles, but, paradoxically, these muscles are virtually lacking in segmental reflexes (Abrahams et al., 1975; Richmond and Loeb, 1992). We do not know how stability is maintained in these muscles. Interestingly, the long neck muscles are notorious sources of disability following idiopathic “strain”.
5. Parallel neuromuscular
compartments
for differential
action
Multiarticular muscles with parallel (nonpennate) fascicles require considerable girth to generate significant amounts of tension and they lack the convergent attachment site provided by an aponeurotic tendon. Large, parallel-fibred muscles such as biceps femoris and tensor fasciae latae have skeletal attachments that span extended lengths of bone, resulting in different skeletal moment arms for fascicles in different regions of the muscle. For many patterns of motion at the various joints, these fascicles experience different amounts and velocity of stretch; some parts of the muscle may do negative work (eccentric, lengthening motion) while others do positive work (concentric, shortening motion). The mechanical characteristics of muscle and the types of fusimotor and proprioceptive control needed under these various circumstances are quite different (Loeb, 1985). It is not surprising that such muscles appear commonly to be divided into multiple, parallel neuromuscular compartments, each innervated by a separate population of motoneurons whose motor territories are narrow strips oriented parallel to the length of the muscle (Loeb et al., 1987; Chanaud et al., 1991a,b). EMG activity recorded from different compartments may vary markedly across the width of the muscle (Pratt et al., 1991; Chanaud and Macpherson, 19911, which may confound EMG studies of muscle function that do not record from each of these compartments separately.
554
G.E. Loeb, F.J.R. Richmond/Human
Mouement
Science I3 (1994) 545-556
Paradoxically, the motoneurons that supply differentially recruited compartments may be thoroughly intermingled in the spinal motor nucleus of the whole muscle (Gordon et al., 1991). This contradicts the suggestion that the topography of the motor nuclei is sufficient to direct neural input to the appropriate target motoneurons (Windhorst et al., 1989). Actually, topography must play a limited role even in motor nuclei that are known to be recruited homogeneously because the dendritic trees of motoneurons from the various motor nuclei tend to be diffusely intermingled throughout most of the ventral and intermediate laminae of the spinal cord (Rose and Keirstead, 1988).
6. Conclusions Multiarticular muscles are uniquely suited to the efficient performance of certain kinds of motor tasks, as discussed elsewhere in this issue. Quantitative understanding of their mechanical contributions often entails the development of mathematical models of the relationships among kinematics, kinetics and motoneuronal activation. Many of these muscles seem to have unusually complex architectural forms, whose identification and characterization require more types of data than are generally provided in most morphometric studies. In order to make this problem tractable for biomechanicians, it will be important to identify which of these architectural details actually affect the work of the muscle under normal conditions of use and which only pose control and stability problems that are normally handled by the nervous system without mechanical consequences. The job of the neurophysiologist is to identify how the neural circuitry generates control signals that cope with the dynamics of the musculoskeletal system, as opposed to how it might control a set of idealized torque motors such as those employed in robots (with highly limited success, one might note). Thus, in order to understand sensorimotor neurophysiology, it is necessary to understand musculoskeletal structure and its consequences for the normal work of the muscles. The nervous system is well suited to solving some kinds of control problems (such as those involving accuracy and precision) but not others (such as those requiring speed and mechanical robustness). Some of the control problems that arise in skeletal movement seem already to have been addressed by mechanical properties that are intrinsic to musculoskeletal architecture; other control problems are actually created by
G.E. Loeb, F.J.R. Richmond/Human
Mouement Science 13 (1994) 545-556
555
those mechanical solutions. Thus, the neurophysiologist is faced with a similar problem of separating the neural circuitry related to accomplishing motor tasks from that related to avoiding injury during the performance of such tasks.
Acknowledgements
The support by the Dystrophy Association Excellence for Neural United States National
Canadian Medical Research Council, the Muscular of Canada, the Canadian Network of Centers of Regeneration and Functional Recovery, and the Institutes of Health is gratefully acknowledged.
References Abrahams, V.C., F.J.R. Richmond and P.K. Rose, 1975. Absence of monosynaptic reflex in dorsal neck muscles of the cat. Brain Research 95, 130-131. Burkett, L.N., 1970. Causative factors in hamstring strains. Medicine and Science in Sports and Exercise 2, 39-42. Chanaud, C.M. and J.M. Macpherson, 1991. Functionally complex muscles of the cat hindlimb. III. Differential activation within biceps femoris during postural perturbation. Experimental Brain Research 85, 271-280. Chanaud, CM., C.A. Pratt and G.E. Loeb, 1991a. Functionally complex muscles of the cat hindlimb. II. Mechanical and architectural heterogeneity in the biceps femoris. Experimental Brain Research 85, 257-270. Chanaud, C.M., C.A. Pratt and G.E. Loeb, 1991b. Functionally complex muscles of the cat hindlimb. V. The roles of histochemical fibre-type regionalization and mechanical heterogeneity in differential muscle activation. Experimental Brain Research 85, 300-313. Dornan, P., 1971. A report on 140 hamstring injuries. Australian Journal of Sports Medicine 4, 30-36. English, A.W. and 0. Weeks, 1987. Anatomical and functional analysis of cat biceps femoris and semitendinosus muscles. Journal of Morphology 191, 161-175. Gans, C., 1982. Fibre architecture and muscle function. Exercise and Sport Sciences Reviews 11, 160-207. Gans, C., G.E. Loeb and F. de Vree. 1989. Architecture and consequent physiological properties of the semitendinosus muscle in domestic goats. Journal of Morphology 199, 287-297. Gordon, D.C., G.E. Loeb and F.J.R. Richmond, 1991. Distribution of motoneurons supplying cat sartorius and tensor fasciae latae, demonstrated by retrograde multiple-labelling methods. Journal of Comparative Neurology 304, 357-372. Gordon, D.C. and F.J.R. Richmond, 1991. Distribution of motoneurons supplying dorsal suboccipital and intervertebral muscles in the cat neck. Journal of Comparative Neurology 304, 343-356. Heron, M.I. and F.J.R. Richmond, 1993. In-series fibre architecture in long human muscles. Journal of Morphology 215, l-11. Ibraghimov-Beskrovnaya, O., J.M. Ervasti, C.J. Leveille, C.A. Slaughter, S.W. Sernett and K.P. Campbell, 1992. Primary structure of dystrophin-associated glycoproteins linking dystrophin to the extracellular matrix. Nature (London) 355. 696-702.
5.56
G.E. Loeb, F.J.R. Richmond /Human
Mouement Science 13 (1994) 545-556
Joyce, G.S., P.M.H. Rack and D.R. Westbury, 1969. Mechanical properties of cat soleus muscle during controlled lengthening and shortening movements. Journal of Physiology (London) 204, 461-474. Liemohn, W., 1978. Factors related to hamstring strains. Journal of Sports Medicine 18, 71-76. Loeb, G.E., 1985. Motoneuron task groups - Coping with kinematic heterogeneity. Journal of Experimental Biology 11.5, 137-146. Loeb, G.E., 1989. ‘The functional organization of muscles, motor units, and tasks’. In: M.D. Binder and L.M. Mendell (Eds.), The segmental motor system. New York: Oxford University Press. Loeb, G.E., C.A. Pratt, CM. Chanaud and F.J.R. Richmond, 1987. Distribution and innervation of short, interdigitated muscle fibres in parallel-fibred muscles of the cat hindlimb. Journal of Morphology 191, l-15. Morgan, D.L., 1985. From sarcomeres to whole muscles. Journal of Experimental Biology 115, 69-78. Pratt, C.A., CM. Chanaud and G.E. Loeb, 1986. Single motor unit territories in the cat sartorius. Society of Neuroscience Abstracts 12, 1083. Pratt, C.A., C.M. Chanaud and G.E. Loeb, 1991. Functionally complex muscles of the cat hindlimb. IV. Intramuscular distribution of movement command signals and cutaneous reflexes in broad, bifunctional tight muscles. Experimental Brain Research 85, 281-299. Richmond, F.J.R. and D.A. Bakker, 1982. Anatomical organization and sensory receptor content of soft tissues surrounding upper cervical vertebrae in the cat. Journal of Neurophysiology 48, 49-61. Richmond, F.J.R., G.J. Bakker, D.A. Bakker and M.J. Stacey, 1986. The innervation of tandem muscle spindles in the cat neck. Journal of Comparative Neurology 245, 483-497. Richmond, F.J.R. and G.E. Loeb, 1992. Electromyographic studies of neck muscles in the intact cat: II. Reflexes evoked by muscle nerve stimulation. Experimental Brain Research 88, 59-66. Richmond, F.J.R., D.R.R. MacGillis and D.A. Scott, 1985. Muscle-fibre compartmentalization in cat splenius muscles. Journal of Neurophysiology 53, 868-885. Richmond, F.J.R., D.B. Thomson and G.E. Loeb, 1992. Electromyographic studies of neck muscles in the intact cat. I. Patterns of recruitment underlying posture and movement during natural behaviours. Experimental Brain Research 88, 41-58. Richmond, F.J.R. and P.P. Vidal, 1988. ‘The motor system: Joints and muscles of the neck’. In: B.W. Peterson and F.J.R. Richmond (Eds.), Control of head movement (pp. l-21). New York: Oxford University Press. Rose, P.K. and S.A. Keirstead, 1988. ‘Cervical motoneurons’. In: B.W. Peterson and F.J.R. Richmond (Eds.1, Control of head movement. New York: Oxford University Press. Schwarzacher, V.H.G., 1959, iiber die Lange und Anordnung der Muskelfasern in menschlichen Skeletmuskeln. Acta Anatomica 37, 217-231. Scott, S.H., D.B. Thomson, F.J.R. Richmond and G.E. Loeb, 1992. Neuromuscular organization of feline anterior sartorius: II. Intramuscular length changes and complex length-tension relationships during stimulation of individual nerve branches. Journal of Morphology 213, 171-183. Selbie, W.S., D.B. Thomson and F.J.R. Richmond, 1993. Sub-occipital muscles in the cat neck: Morphometry and histochemistry of the rectus capitis muscle complex. Journal of Morphology 216, 47-63. Smits, E., M.I. Heron, P.K. Rose, T. Gordon and F.J.R. Richmond, 1991. Motor-unit distribution in feline anterior sartorius muscle. Society of Neuroscience Abstracts 17, 1393. Thomson, D.B., S.H. Scott and F.J.R. Richmond, 1991. Neuromuscular organization of feline anterior sartorius: I. Asymmetric distribution of motor units. Journal of Morphology 210, 147-162. Trotter, J.A., 1990. Interfibre tension transmission in series-fibred muscles of the cat hindlimb. Journal of Morphology 206, 351-361. Windhorst, U., T.M. Hamm, D.G. Stuart, 1989. On the function of muscle and reflex partitioning. Behavioral and Brain Sciences 12, 629-681. Young, R.P., S.H. Scott and G.E. Loeb, 1993. The distal hindlimb musculature of the cat; Multiaxis moment arms of the ankle joint. Experimental Brain Research 96, 141-151.
