INVITED REVIEW - Freee.guigon.free.fr/rsc/article/LieberFriden00.pdf · INVITED REVIEW ABSTRACT: Skeletal muscle architecture is the structural property of ... “the arrangement

INVITED REVIEW ABSTRACT: Skeletal muscle architecture is the structural property ofwhole muscles that dominates their function. This review describes the basicarchitectural properties of human upper and lower extremity muscles. Thedesigns of various muscle groups in humans and other species are analyzedfrom the point of view of optimizing function. Muscle fiber arrangement andmotor unit arrangement is discussed in terms of the control of movement.Finally, the ability of muscles to change their architecture in response toimmobilization, eccentric exercise, and surgical tendon transfer is reviewed.Future integrative physiological studies will provide insights into the mecha-nisms by which such adaptations occur. It is likely that muscle fibers trans-duce both stress and strain and respond by modifying sarcomere number ina way more suited to the new biomechanical environment.

© 2000 John Wiley & Sons, Inc. Muscle Nerve 23: 1647–1666, 2000

FUNCTIONAL AND CLINICAL SIGNIFICANCEOF SKELETAL MUSCLE ARCHITECTURE

RICHARD L. LIEBER, PhD, 1,2 and JAN FRIDE N, MD, PhD3

1 Veterans Affairs Medical Center and University of California, Department ofOrthopaedics, U.C. San Diego School of Medicine and V.A. Medical Center,3350 La Jolla Village Drive, San Diego, California 92161, USA2 Veterans Affairs Medical Center and University of California, Department ofBioengineering, San Diego, California, USA3 Department of Hand Surgery, Goteborg University, Goteborg, Sweden

Accepted 5 June 2000

Structure–function relationships in skeletal musclehave been described and examined for over a cen-tury. Classic studies have elucidated the microscopicand ultrastructural properties of skeletal muscle fi-bers, yielding great insights into their function. How-ever, less attention has been paid to the studies ofthe macroscopic properties of skeletal muscle tissuesdating back to the 1600s (see discussion and refer-ences in Kardel27). This macroscopic arrangementof muscle fibers is known as a muscle’s architec-ture.20 Because muscle architecture is a primary de-terminant of muscle function, understanding thisstructure–function relationship is of great practicalimportance. This understanding not only clarifiesthe physiological basis of force production andmovement but also provides a scientific rationale forsurgery that may involve tendon-transfer procedures,

provides guidelines for electrode placement duringelectromyographic measures of muscle activity, ex-plains the mechanical basis of muscle injury duringnormal movement, and aids in the interpretation ofhistological specimens obtained from muscle biop-sies. The purpose of this review is to describe thetheoretical significance of muscle architecture, todescribe the basic architectural properties of humanupper and lower extremity muscles, and to intro-duce advanced topics that represent current issuesnot yet resolved in the literature.

BASIC ARCHITECTURAL DEFINITIONS

It is well-known that skeletal muscle is highly orga-nized at the microscopic level, as witnessed by theincredible number and diversity of electron micro-graphs and schematics of muscle sarcomeres thathave been published in review articles and text-books. However, with few exceptions, the arrange-ment of muscle fibers within and between muscleshas received much less attention. Muscle fibers areoften depicted as projecting in bundles (fascicles)from an origin on a proximal tendon plate to aninsertion more distally. This simply does not do jus-tice to the wide array of muscle “designs” that areapparent throughout the animal kingdom. The ar-

Abbreviations: CT, computed tomography; dSL/dv, sarcomere lengthchange during joint rotation; ECRB, extensor carpi radialis brevis; ECRL,extensor carpi radialis longus; EMG, electromyography; Lf, muscle fiberlength; Lm, muscle length; Lo, optimal length; MRI, magnetic resonanceimaging; MTU, muscle tendon unit; PCSA, physiological cross-sectionalarea; Po, maximum tetanic tension; r, muscle density; r, moment arm;ROM, range of motion; u, muscle fiber pennation angle; Vmax, maximumcontraction velocityKey words: architecture; muscle design; muscle fiber; skeletal muscle;tendon transfer surgeryCorrespondence to: R.L. Lieber; e-mail: [email protected]

© 2000 John Wiley & Sons, Inc.

Skeletal Muscle Architecture MUSCLE & NERVE November 2000 1647

chitecture of a given muscle is extremely consistentbetween individuals of the same species, giving riseto the concept that certain constraints that deter-mine muscle architectural properties are pres-ent.5,26,34,36,41,55,58,59,65 Although much attention hasbeen paid to factors such as fiber-type distribution indetermining muscle function, there is no questionthat function is also strongly determined by a mus-cle’s architecture.7

Skeletal muscle architecture can be defined as“the arrangement of muscle fibers within a musclerelative to the axis of force generation.”31 Whereasmuscle fibers have a relatively consistent fiber diam-eter between muscles of different sizes and fiber sizeis directly proportional to fiber force generation, ar-chitectural differences between muscles demon-strate much more variability and more strongly affectfunction. Although it might be proposed that differ-ent muscles produce different amounts of forcebased on differences in fiber size, fiber size betweenmuscles actually varies little. Architectural differ-ences between muscles are the best predictors offorce generation. The various types of architecturalarrangements are as numerous as the number ofmuscles themselves. For discussion purposes, how-ever, we describe three general classes of muscle fi-ber architecture.

Muscles composed of fibers that extend parallelto the muscle’s force-generating axis are describedas having a parallel or longitudinal architecture (Fig.1A). Although the fibers may project along a lineparallel to the force-generating axis, experimentalstudies of mammalian muscle suggest that musclefibers do not extend the entire muscle length. Infact, detailed studies of muscle fiber lengths demon-strate that muscle fibers may not even extend theentire length of a fascicle.41,47 Muscles with fibersthat are oriented at a single angle relative to theforce-generating axis are described as having uni-pennate architecture (Fig. 1B). The angle betweenthe fiber and the force-generating axis has beenmeasured at resting length in mammalian muscles ofvery different designs and varies from about 0° to30°. It becomes obvious when performing muscledissections that most muscles fall into the third andmost general category, multipennate muscles—muscles constructed of fibers that are oriented atseveral angles relative to the axis of force generation(Fig. 1C). Although these three designations areoversimplified, they provide a vocabulary with whichto describe muscle designs. Because fiber orienta-tion may have nothing to do with classic anatomicalaxes, determination of muscle architecture is impos-sible from a single biopsy or even a magnetic reso-

FIGURE 1. Artist’s conception of three general types of skeletal muscle architecture. (A) Longitudinal architecture in which muscle fibersrun parallel to the muscle’s force-generating axis, as in the biceps brachii. (B) Unipennate architecture in which muscle fibers run at a fixedangle relative to the muscle’s force-generating axis, as in the vastus lateralis muscle. (C) Multipennate architecture in which muscle fibersrun at several angles relative to the muscle’s force-generating axis (gluteus medius muscle). Lf, muscle fiber length; Lm, muscle length.

1648 Skeletal Muscle Architecture MUSCLE & NERVE November 2000

nance imaging (MRI), computerized tomography(CT), or ultrasound image, as these methods cannotaccount for variations in fiber length and orientationthat occur along the muscle length. Thus, experi-mental methods have been developed to character-ize the architectural properties of skeletal muscle.

EXPERIMENTAL DETERMINATION OF SKELETALMUSCLE ARCHITECTURE

Quantitative studies of muscle architecture were pio-neered by Gans and Bock20 and Gans and De Vries,21

who developed precise methodology for definingmuscle architecture based on microdissection ofwhole muscles. The usual parameters included inan architectural analysis are muscle length (Lm),fiber length (Lf), pennation angle (i.e., the fiberangle relative to the force-generating axis, u), andphysiological cross-sectional area (PCSA). Typically,muscles are chemically fixed in 10% buffered forma-lin in order to maintain fiber integrity during dissec-tion. The muscles should be chemically fixed whileattached to the skeleton, to roughly preserve theirphysiological length, or physiological length in theskeleton should at least be noted. After fixation,muscles are dissected from the skeleton, their massdetermined, and their pennation angle and musclelength are measured.

The majority of reported studies fix muscles bydirect immersion into fixative, relying on diffusionof the fixative throughout the tissue thickness to fa-cilitate subsequent dissection. However, for verythick muscles, the quality of the resulting fixationmay not be adequate for architectural measurementsto be made. Some investigators advocate allowingthe muscle to go into rigor prior to fixation to pre-serve sarcomere length homogeneity. Recently, weimplemented a high-pressure infusion method tochemically fix large human muscles of the shoulderand chest.17 Muscles from fresh cadaveric specimenswere fixed by high-pressure perfusion of 10% buff-ered formaldehyde via the carotid artery. To achieveadequate pressure, ∼50 L containers filled with fixa-tive were placed at a height of approximately 3 mabove the cadaver. The infusion tubing had a diam-eter of approximately 5 mm. The carotid artery wascannulated and fixative introduced under pressureover a period of about 3 days. This method producedthe best fixation of muscle tissue from cadavericspecimens that we have observed in approximately10 years of this type of experimentation.

After fixation, u is measured by determining theaverage angle of the fibers on the superficial musclesurface relative to the muscle’s axis of force genera-tion. Of course, this is only an estimate, because

angles may vary from superficial to deep and evenfrom proximal to distal (Fig. 1). Although more so-phisticated methods could be developed for mea-surement of pennation angle, it is doubtful theywould provide a great deal more insight into musclefunction, because variations in pennation angle maynot strongly affect function20 (see below).

Muscle length Lm is defined as “the distance fromthe origin of the most proximal muscle fibers to theinsertion of the most distal fibers.”31 This is not thesame as Lf because of the variable degree of “stagger”seen in muscle fibers as they arise from and insertonto tendon plates. To date, muscle fiber length canbe determined only by microdissection of individualfibers from fixed tissues or by laborious identifica-tion of fibers by glycogen depletion on serial sectionsalong the length of the muscle.47 Unless investiga-tors are explicit, when they refer to muscle fiberlength, they are probably referring to muscle fiberbundle length, because it is extremely difficult toisolate intact fibers, which run from origin to inser-tion, especially in mammalian tissue.41,47 Thus, whenmicrodissection is performed, bundles consisting of5–50 fibers are typically used to estimate Lf. Becauseof the many different methods available for Lf deter-mination, a distinction between “anatomical” fiberlength and “functional” fiber length may be made.The anatomical fiber length determined by micro-dissection may not accurately represent the functionof the fibers within the muscle, because fibers maybe broken and their activation pattern unknown. Incontrast, when determining Lf by glycogen depletionof single motor units,47 it is clear that the entire fiberlength is activated and can be identified histologi-cally.

The final experimental step required to performarchitectural analysis of a whole muscle is to measuresarcomere length within the isolated fibers. This isnecessary to compensate for differences in musclelength that occur during fixation. In other words, toconclude that a muscle has “long fibers,” one mustensure that it truly has long fibers and not that itwas fixed in a highly stretched position correspond-ing to a long sarcomere length. Similarly, muscleswith “short fibers” must be further investigated toensure that they were not simply fixed at a shortsarcomere length. In order to permit such conclu-sions, architectural measurements should be nor-malized to a constant sarcomere length, which elimi-nates fiber length variability due to variation infixation length. This provides a reference value thatcan even be related back to the physiological lengthif the relationship between muscle length and joint


position is noted. Then, based on measured archi-tectural parameters and joint properties, the rela-tionship between sarcomere length and joint anglecan be calculated. Because sarcomere length alsostrongly influences muscle force generation, an un-derstanding of the relationship between sarcomerelength change and movement has been used inmany studies to provide added understanding ofmuscle design.8,37,38,52,54

After measurement of the architectural param-eters described above, the PCSA is calculated. ThePCSA of a muscle is the only architectural parameterthat is directly proportional to the maximum tetanictension generated by the muscle. This value is almostnever the actual cross-sectional area of the muscle asmeasured in any of the traditional anatomicalplanes, as would be obtained, for example, using anoninvasive imaging method such as MRI, CT, orultrasound. Theoretically, PCSA represents the sumof the cross-sectional areas of all the muscle fiberswithin the muscle. It is calculated using eq. 1, whichwas verified experimentally by Powell et al.,50

PCSA (mm2! =muscle mass (g) ? cosine u

r ~g/mm3! ? fiber length (mm!(1)

where r represents muscle density (1.056 g/cm3 formammalian muscle) and u represents surface pen-nation angle.

Equation 1 represents muscle volume (mass/density) divided by fiber length and has units of area(in this case, cm2). Because fibers may be oriented atan angle relative to the axis of force generation, thecosine term is often included, considering that notall the fiber tensile force is transmitted to the ten-dons. This idea can be envisioned using a model ofa muscle fiber pulling with x units of force at anangle u relative to the muscle axis of force genera-tion. In this configuration, only a component ofmuscle fiber force will actually be transmitted alongthe muscle axis, which will be x ? cosu. In otherwords, in this scenario, pennation is seen to result ina loss of muscle force relative to a muscle with thesame mass and fiber length but with zero pennationangle.

EXPERIMENTAL VERIFICATION OF MUSCLEARCHITECTURE EQUATIONS

Cross-Sectional Area. The accuracy of Eq. 1 washighlighted by experimental comparison betweenthe calculated maximum muscle tetanic tension(based on PCSA) and measured maximum tetanictension (using traditional physiological testing tech-

niques) of guinea pig skeletal muscle.50 It was foundthat the estimations and predictions agreed withinexperimental error and that the only exception wasthe soleus muscle (Fig. 2). For all “fast” muscles,force per unit cross-sectional area was ∼22.5N/cm2,which serves as a nominal value of specific tensionfor mammalian muscle.

A potential problem with eq. 1 is that it assumesmuscle fiber pennation angle is constant duringmuscle contraction. Experimental measurement ofmuscle fiber pennation angle in the unipennate ratgastrocnemius muscle have revealed that this is notthe case.70 Zuurbier and Huijing69 placed small wiremarkers across the gastrocnemius surface and filmedmarker movement during muscle contraction. Usingthese data, the authors measured fiber pennationangle and showed that muscle fiber angle varied con-siderably as muscle length was altered throughoutthe normal physiological range of motion (Fig. 3A).While resting fiber u was approximately 30°, duringisotonic contraction, u increased to almost 60° andthe angle between the muscle aponeurosis and themuscle axis rotated from about 10° to 15°. The factthat muscle fibers appear to be free to rotate duringcontraction has a number of implications. First,PCSA as calculated from eq. 1 may be somewhatarbitrary in its ability to predict force. Second, fiber

FIGURE 2. Comparison between measured Po and estimated Po,based on measurement of PCSA in guinea pig muscles. Notethat PCSA and Po are proportional for all muscles except thesoleus, the one muscle with a preponderance of slow musclefibers. (Abbreviations: EDL, extensor digitorum longus; FHL,flexor hallucis longus; LG, lateral gastrocnemius; MG, medialgastrocnemius; PLT, plantaris; SOL, soleus; TA, tibial anterior;TA–TAA, combined tibialis anterior accessorius and tibialis ante-rior; TAA, tibialis anterior accessorius; TP, tibialis posterior.) Notethe close linear correlation between PCSA and measured Po.The slope of the line 22.5N/cm2 represents a reasonable valuefor muscle-specific tension. (Data replotted from Powell et al.50)


rotation has a significant impact on the estimation ofmuscle fiber contraction velocity based on the mea-surement of whole muscle velocity. During shorten-ing of the rat gastrocnemius, if fiber contractionvelocity is calculated holding pennation angle con-stant, it will simply scale with whole muscle velocity.However, if fiber rotation is included, fiber velocity isactually much lower than whole muscle velocity,which, functionally, results in prediction of greaterforce generation by the muscle fibers (Fig. 3B).

Thus, fiber rotation during muscle shortening pro-vides a degree of freedom that allows muscle fibersto maintain a higher level of force generation thanwhen fibers are constrained to maintain a constantpennation angle.

Does significant fiber rotation occur in humanmuscles? An affirmative answer to this question wasprovided in a most dramatic example by Fukunagaand colleagues19 and Kawakami and colleagues,28

who measured fascicle plane orientation during vol-untary contraction of the human quadriceps, dorsi-flexor, and plantarflexor muscles. Their measure-ments revealed that during low-level voluntarycontraction, vastus lateralis pennation angle in-creased from 14° with the knee extended to 21° withthe knee flexed, and, during this voluntary contrac-tion, fascicle length decreased from 126 mm to 67mm. Similarly, the medial gastrocnemius increasedfrom about 20° to 45°, with a similar significant de-gree of shortening (Fig. 4). These data provide areal-time picture indicating that muscle fiber short-ening and rotation are simultaneous and normalevents that occur during muscle contraction. Fiberrotation during muscle contraction permits tensileforce transmission to occur even when muscle fibersare oriented at an angle relative to the muscle’sforce-generating axis. The fact that pennation anglesare small at muscle resting lengths (0–30°) probablyaccounts for the agreement between experimentand theory observed by Powell et al.50 Thus, in pre-dicting maximum muscle force–producing capacity,correction for fiber angulation may not be necessary.

Muscle Fiber Length. Even though it is often statedthat muscle fiber length is proportional to fiber ex-cursion (or velocity), there has not been a compre-hensive study in mammalian muscle, analogous tothe study described above for PCSA that confirmsthis relationship quantitatively. However, there is agood deal of experimental evidence available in theliterature to suggest that this relationship is valid.First, in mechanical studies of isolated frog singlemuscle fibers, in which fiber length, and thus thenumber of sarcomeres in series, is easily measured,maximum contraction velocity is directly propor-tional to fiber length, as is the width of the isometriclength–tension relationship.15,67 This is the reasonthat muscle contraction velocities are often normal-ized and expressed in “fiber length/s” or “sarcomerelengths/s.” Second, in a mechanical and anatomicalstudy of the cat semitendinosus muscle, which rep-resents a unique model in that it is composed ofdistinct proximal and distal heads separated by a ten-dinous inscription and which has distinct innerva-

FIGURE 3. Changes in muscle structural properties as a functionof muscle length in rat medial gastrocnemius muscle. (A) Fiberpennation angle. This angle increases as the muscle shortens.(B) Fiber velocity expressed relative to muscle velocity duringisotonic shortening. Notice that at shorter lengths, fiber velocity isa decreasing percentage of muscle velocity, due to fiber rotation.(C) Aponeurosis angle. (Data replotted from Zuurbier and Hui-jing.70)


tion to each head, the maximum contraction velocityof the two heads stimulated simultaneously was thesame as the sum of the maximum contraction veloc-ity of the two heads stimulated individually.2

The relationship between fiber length and wholemuscle properties is also complicated by the obser-vation that some very long feline muscles are com-posed of relatively short muscle fibers arranged inseries.41 The cat sartorius, tenuissimus, and semiten-dinosus muscles were microdissected and shown tobe composed of relatively short fibers, 2–3 cm inlength, that were arranged in series. By stainingthese specimens with acetylcholinesterase, the au-thors also showed that the endplates occurred inabout the same serial arrangement. These data werecombined with electromyographic data that demon-strated that the “in series” fibers were innervated bybranches of axons also arranged in series, so that all

serial fibers would be expected to be activated simul-taneously. This led to the suggestion that short fibersarranged in series are simultaneously activated to acteffectively like long muscle fibers. The reason thiswas viewed as a good design is that it is known thatelectrical conduction along muscle fibers is relativelyslow (2–10 m/s).13 Because of the long length ofthese cat muscles (10–15 cm), it was conceived thatby the time the action potential reached the end ofthe fiber, the central region might already be relax-ing. Mechanically, this is clearly unfavorable. This isnot to say that longer fibers do not exist within catmuscles. In a histological study of the cat tibialis an-terior, Ounjian et al.47 traced single muscle fiberswithin single motor units along the muscle bellycross-section and clearly demonstrated that the fi-bers did not run the entire muscle length nor eventhe entire fascicle length (Fig. 5). Fibers within mo-tor units ranged from ∼8 mm to ∼50 mm. Further-more, the authors elucidated the general patternthat slow (S) units tended to have more uniformfiber lengths compared with fast (F) units. Fiberends within the F units also tapered to a greaterextent compared with S units. Additionally, someunits actually began and ended within a fascicle,whereas others extended all or part of the length ofa fascicle (Fig. 5). These data, taken in concert withthe motor unit stiffness data of Petit et al.48 and Petitet al.,49 imply that the higher stiffness of the S com-pared with F units could result from the relativelyshort muscle fibers within those units and could rep-resent a beneficial design for postural control. Be-cause the S units are implicated in posture and alsohave a relatively high stiffness, small positional per-turbations would be effectively damped by the intrin-sic properties of the fibers within those units. Itwould also be predicted that slow units would have anarrower length–tension relationship comparedwith fast units as a direct reflection of differences inmuscle fiber length, but this suggestion has only re-ceived weak experimental support.25

To summarize the results of muscle fiber lengthdeterminations, most studies report fiber bundlelengths rather than fiber lengths. These may be ac-tual anatomical lengths based on microdissection ora functional length based on a physiological phe-nomenon such as glycogen depletion of single fi-bers. However, because fibers may terminate end toend, functionally, a fascicle may perform like a singlemuscle fiber of equivalent total length. Fibers mayterminate within the muscle belly in the complexextracellular matrix composed of endomysial con-nective tissue.62 These connective tissues merge intoa final “external tendon” where force can be applied

FIGURE 4. Ultrasound image of human medial gastrocnemiusmeasured in the sagittal plane. (A) Muscle image at rest demon-strating passive fiber angle (up) of 20° relative to the aponeurosis.(B) Muscle image with the muscle activated demonstrating activefiber angle (ua) of 45° relative to the aponeurosis. (Ultrasoundimages courtesy of T. Fukanaga and Y. Kawakami.)


to bone. It would be important to demonstrate quan-titatively the relationship between fiber or fasciclelength and measured maximum contraction velocityof muscle (Vmax).

ARCHITECTURE OF HUMAN SKELETAL MUSCLES

Several architectural investigations have been per-formed in human upper and lower limb mus-cles.5,18,34,36,65 Reported pennation angles normallyrange from 0° to 30°, and the ratio of muscle fiberlength to muscle length ranges from ∼0.2 to ∼0.6. Inother words, even the most “longitudinally oriented”muscles have fiber bundles that extend only about60% of the muscle length.

Muscles of the Human Lower Limb. The two mostimportant muscle architectural parameters aremuscle PCSA (proportional to maximum muscleforce) and muscle fiber length (proportional to

maximum muscle excursion). The relationship be-tween these two parameters for human muscles isshown in Figure 6. Although each muscle is uniquein terms of its architecture, taken as functionalgroups (e.g., hamstrings, quadriceps, dorsiflexors,plantarflexors), a number of generalizations can bemade regarding lower extremity muscles. Quadri-ceps muscles are characterized by their relativelyhigh pennation angles, large PCSAs, and short fi-bers. In terms of design, these muscles appear suitedfor the generation of large forces. The hamstrings,on the other hand, by virtue of their relatively longfibers and intermediate PCSAs, appear to be de-signed for large excursions. Specifically, note thatthe sartorius, semitendinosus, and gracilis muscleshave extremely long fiber lengths and low PCSAs,which permit large excursions at low forces (Fig. 6).Specialization of functional groups also appears tobe true of the plantarflexors and dorsiflexors. A verygeneral conclusion might be that the antigravity ex-tensors are designed more toward force production,and the flexors are designed more for long excur-sions. The most extreme example of such a design isthe soleus muscle, with its high PCSA and short fiberlength, suitable for generating high force with smallexcursion. This would suit it well for a postural sta-bilization role.

The relatively high pennation angles of the quad-riceps also have implications for muscle biopsy analy-sis and electromyogram (EMG) measurement. Be-cause the fibers are relatively short, a biopsy obtainedalong the length of the muscle may not be represen-tative of all the fibers along the muscle, which arestaggered to a high degree. Experimental quantifi-cation of repeated biopsies from four rhesus mon-keys was studied in three different locations alongthe length of the soleus, medial gastrocnemius, andtibialis anterior muscles.56 The authors demon-strated a much greater fiber-type percentage variabil-ity between animal subjects than between biopsieswithin a subject. The authors also demonstratedthat, in spite of sampling different fibers alongmuscles with very different architectures (Lf/Lm ra-tios ranging from 0.23 to 0.35), it was possible toobtain representative percentages within these dif-ferent muscle regions. Fiber-type variability was5–10% within a muscle but as much as 30% betweenthe same muscles of different animal subjects.

With regard to EMG measurements, a similarsampling problem may arise as a result of musclefiber stagger. Electrodes placed in one region of themuscle may not provide an electrical signal that is

FIGURE 5. (A) Muscle fiber length and position determined byglycogen depletion within seven different single motor units of thecat tibialis anterior muscle. Motor units 1–5 were physiologicallytyped as fast, and motor units 6 and 7 were physiologically typedas slow. The most proximal end of the muscle is defined as 0 mm,whereas the most distal portion of the muscle is denoted by the⊥ symbol. Note that some muscle fibers actually begin and endwithin the muscle fascicle itself. (B) Cross-sectional area of fiberswithin motor unit 4 shown in A. Note that, typically, fibers taper inending within the fascicle. (Figure redrawn from Ounjian et al.47)


representative of motor units from different regionsof the same muscle.40 This is due to two factors. Firstand most obvious, muscle fibers do not extend thelength of the muscle, and second, a natural grada-tion exists in fiber-type percentage and thus motorunit types from superficial to deep within a muscle.6

Because motor units are activated in a stereotypicalfashion from slow to fast,23 this may affect durationand amplitude of EMG signals measured at differentdepths. It should be noted, however, that the extentto which this inability to sample uniformly affectseither clinical judgment or the understanding ofmuscle activation has not been clearly determined.A second level of complexity that may affect theextent to which an EMG signal is representative ofmuscle function arises from the fact that somemuscles, such as the cat lateral gastrocnemius, dem-onstrate compartmentalization.16,64 Under theseconditions, separate portions of muscles with unique

fiber-type distributions are innervated by distinctlydifferent motor nerves. As a result, their activationpattern and general level of use can differ, in spite ofthe fact that they are in the same muscle.

Muscles of the Human Upper Limb. In light of thespecialization observed in the lower limb, it is prob-ably not surprising to note that there is also a highdegree of specialization “built into” upper extremitymuscles, by virtue of their architecture. Such special-ization makes sense in light of the fact that a greatdeal more functional diversity is seen in upper ex-tremity compared with lower extremity movement.From an architectural point of view, the superficialand deep digital flexors are similar to each other butare different from the digital extensors (Fig. 7). Theflexors would be predicted to generate almost twicethe force as the extensors and with a slightly greateractive range. As another example, based on its very

FIGURE 6. Scatter graph of the fiber length and cross-sectional areas of muscles in the human lower limb. Fiber length is proportionalto muscle excursion, and cross-sectional area is proportional to maximum muscle force. Thus, this graph can be used to compare therelative forces and excursions of muscles within the lower limb. (Abbreviations: AB, adductor brevis; AL, adductor longus; AM, adductormagnus; BFl, biceps femoris, long head; BFs, biceps femoris, short head; EDL, extensor digitorum longus; EHL, extensor hallucis longus;FDL, flexor digitorum longus; FHL, flexor hallucis longus; GR, gracilis; LG, lateral gastrocnemius; MG, medial gastrocnemius; PB,peroneus brevis; PEC, pectineus; PL, peroneus longus; PLT, plantaris; POP, popliteus; RF, rectus femoris; SAR, sartorius; SM, semi-membranosus; SOL, soleus; ST, semitendinosus; TA, tibialis anterior; TP, tibialis posterior; VI, vastus intermedius; VL, vastus lateralis;VM, vastus medialis.) (Data from Wickiewicz et al.65)


high PCSA, the flexor carpi ulnaris is expected togenerate very high forces. Examination of this typeof information can be used to compare functionalproperties between muscles that might be surgicallytransferred to restore lost function (see below). In-tuitively, it might be considered important to matchthe transferred muscle’s architectural properties tothe architectural properties of the muscle whosefunction was lost.

SIGNIFICANCE OF MUSCLE ARCHITECTURE INSURGICAL TENDON TRANSFER

In addition to improving our understanding ofmuscle anatomy and function, elucidation of musclearchitecture may ultimately provide information use-ful for selection of muscles used in tendon transfers.To substitute lost muscle function, the distal tendonsof muscles are often transferred from one position toanother,3,4,22,51 a procedure known as a “tendontransfer.” It would seem reasonable to select a donormuscle with architectural properties that are similar

to the original muscle in order to perform the origi-nal muscle’s function. (Numerous other factors alsoinfluence donor selection, including donor muscleavailability, donor muscle morbidity, preoperativestrength, integrity, expendability, synergism, transferroute and direction, and surgical experience andpreference.)

Surgical Restoration of Digital Extension. We envi-sion that architectural differences might be useful intendon transfer when making a choice involvingmultiple donors or when a combination of transfersis available for selection. For example, in the surgicalrestoration of digital extension following high radialnerve palsy, described and accepted potential donormuscles (which are transferred to the extensor digi-torum longus) include the flexor carpi radialis, theflexor carpi ulnaris, the flexor digitorum superfi-cialis to the middle finger, and the flexor digitorumsuperficialis to the ring finger. From the standpoint

FIGURE 7. Scatter graph of the fiber length and cross-sectional areas of muscles in the human forearm. Fiber length is proportional tomuscle excursion, and cross-sectional area is proportional to maximum muscle force. Thus, this graph can be used to compare therelative forces and excursions of arm and forearm muscles. (Abbreviations: BR, brachioradialis; ECRB, extensor carpi radialis brevis;ECRL, extensor carpi radialis longus; ECU, extensor carpi ulnaris; EDC I, EDC M, EDC R, and EDC S, extensor digitorum communis tothe index, middle, ring, and small fingers, respectively; EDQ, extensor digiti quinti; EIP, extensor indicis proprius; EPL, extensor pollicislongus; FCR, flexor carpi radialis; FCU, flexor carpi ulnaris; FDP I, FDP M, FDP R, and FDP S, flexor digitorum profundus muscles; FDSI, FDS M, FDS R, and FDS S, flexor digitorum superficialis muscles; FDS I (C), the combined properties of the two bellies as if they werea single muscle; FDS I (P) and FDS I (D), proximal and distal bellies of the FDS I; FPL, flexor pollicis longus; PQ, pronator quadratus;PS, palmaris longus; PT, pronator teres.) (Data from Lieber et al.34,36)


of architecture alone, the flexor digitorum superfi-cialis to the middle finger most closely resembles theextensor digitorum communis in terms of force gen-eration (i.e., cross-sectional area) and excursion(i.e., fiber length). This is emphasized by the rela-tively close position in “architectural space” of theflexor digitorum superficialis to the middle finger tothe extensor digitorum communis (Fig. 7). If indi-vidual architectural properties are compared, it isclear that the flexor digitorum superficialis to themiddle finger has more than enough excursion com-pared with the extensor digitorum communis,whereas the flexor carpi ulnaris has excessive force-generating potential but may lack sufficient excur-sion. Thus, if the concern were sufficient force, theflexor carpi ulnaris might be chosen, whereas if theconcern were excursion, the flexor digitorum super-ficialis to the middle finger might be chosen. Eitherway, a knowledge of muscle architecture permits aninformed decision to be made. It should be notedthat architectural mismatch between the flexor carpiulnaris and extensor digitorum communis has beenblamed for the poor clinical result of this transfer.39

Surgical Restoration of Thumb Extension. To re-store thumb extensor function in high radial nervepalsy, potential donors include the flexor digitorumsuperficialis to the middle finger, flexor digitorumsuperficialis to the small finger, and palmaris longus.Again, in terms of architecture, the flexor digitorumsuperficialis to the small finger and the palmaris lon-gus are similar to the extensor pollicis longus andtherefore should provide the force generation andexcursion required to restore lost function (Fig. 7).

Surgical Restoration of Thumb Flexion. As a finalexample, following high median nerve palsy, ante-rior interosseous nerve injury, or isolated, irrepa-rable flexor pollicis longus muscle injury, multiplepotential donors for transfer to restore thumb flex-ion are available. These donors include the brachio-radialis, extensor carpi radialis brevis, extensor carpiradialis longus, extensor carpi ulnaris, extensor di-giti quinti, and flexor digitorum superficialis to thering finger. From an architectural standpoint, theextensor carpi radialis brevis, flexor digitorum su-perficialis to the ring finger, and extensor carpi ul-naris are most similar to the flexor pollicis longus(Fig. 7).

Quantification of Architectural Differences betweenPairs of Muscles. To perform a quantitative com-parison between muscles in terms of their architec-tural properties, a method was developed to determine

which architectural properties best discriminated be-tween functional groups and then a scaled calcula-tion was made comparing pairs of muscles usingthese parameters.33 The multivariate method of thediscriminant analysis first determined architecturalproperties which uniquely characterized musclegroups in either the rabbit hindlimb32 or the humanwrist and hand musculature.34,36 For both species, itwas possible to identify five specific discriminatingparameters that successfully characterized the appro-priate functional groups. The best discriminatorswere (in order): Lf, PCSA, Lm, Lf/Lm ratio, andmuscle mass. These five parameters were then usedto calculate an “architectural difference index” forpairs of muscles, according to a modification of thedistance formula for two points in Cartesian coordi-nates. The difference is calculated from linear com-binations of discriminating parameters in order toobtain a single number, the magnitude of which de-scribes the architectural difference between twomuscles (high values implying architectural differ-ences).

SIGNIFICANCE OF MUSCLE ARCHITECTURE INMOTOR CONTROL

One consequence of the architectural specializationnoted throughout the body is that the neuromuscu-lar system is not the only means available to modifymuscular force and excursion. Although neural in-put to muscles can change muscle force, the effec-tiveness of neural input is increased, because indi-vidual muscles are intrinsically designed for aspecific function, large excursion, for example. Thenervous system provides the input signal to modulatethe timing and intensity of the activation signal tothe muscle. It is as if the nervous system acts as thecentral control and the muscle interprets the controlsignal into the actual external action by virtue of itsintrinsic design.

MECHANICAL PROPERTIES OF MUSCLES WITHDIFFERENT ARCHITECTURES

As stated above, muscle force is proportional toPCSA and muscle velocity is proportional to fiberlength. By stating that velocity is proportional to fi-ber length, it is implicit that the total excursion (ac-tive range) of a muscle is also proportional to fiberlength.

Comparison of Two Muscles with Different PCSA.Suppose that two muscles had identical fiber lengthsand pennation angles but one muscle had twice themass (equivalent to saying that one muscle had twicethe number of fibers and thus twice the PCSA). Also


suppose that their fiber-type distributions are iden-tical and that they generate the same force per unitarea (specific tension). The functional difference be-tween these two muscles is shown in Fig. 8. Themuscle with twice the PCSA has a length–tensioncurve with the same shape but is amplified upwardby a factor of two. Thus, the maximum tetanic ten-sion (Po) of the larger muscle will be twice that of thesmaller muscle. Similarly, comparison of schematicforce–velocity curves indicates that the differencesbetween muscles are simply an upward shift in Po

for the larger muscle but the curve retains the sameshape. If both curves were plotted on relative scales(i.e., percent maximum tension instead of absolutetension), the two muscles would appear to have iden-

tical properties. This demonstrates that although ar-chitectural properties profoundly affect the extrinsicmuscle properties (i.e., the properties that vary withabsolute muscle size, such as Po, PCSA, or mass),they have no effect on its intrinsic properties (i.e.,the properties that are independent of absolutemuscle size, such as the shape of the length–tensionor force–velocity curves or fiber length/musclelength ratio).

Comparison of Two Muscles with Different FiberLengths. If two muscles have identical PCSAs andpennation angles but different fiber lengths, thenthe effect of increased fiber length is to increasemuscle velocity (or muscle excursion), as demon-strated in the schematic in Figure 9. Peak absoluteforce of the length–tension curves is identical be-tween muscles, but the absolute muscle active rangeis different. For the same reason that an increasedfiber length increases active muscle range of thelength–tension relationship, it results in an increasein the muscle’s Vmax. As in the example with alteredPCSA, an increase in fiber length increases absoluteproperties but has no effect on the intrinsic proper-ties of the muscle. Experimental support for thisconcept was obtained indirectly in the study men-tioned above on the cat semitendinosus muscle.2

When only the proximal semitendinosus head wasactivated, its Vmax was 224 mm/s, whereas when onlythe distal semitendinosus head was activated, its Vmax

was 424 mm/s. Then, when both heads were acti-vated simultaneously, the whole muscle Vmax was 624mm/s, or the sum of the two velocities. The valuesfor Vmax were proportional to the different lengthsof the proximal and distal heads. These data indicatethat the longer are the fibers in series (equivalent ofsaying the greater number of sarcomeres in series),the greater is the muscle contraction velocity. As ex-pected, maximum isometric tension was essentiallythe same regardless of which activation pattern wasused.

RANGE OF MOTION (ROM) AS A FUNCTIONOF ARCHITECTURE

Muscles with longer fibers have longer functionalranges than do muscles with shorter fibers. This doesnot necessarily indicate that muscles with longer fi-bers are associated with joints that have larger rangesof motion. It is true that a muscle with longer fibersdoes have a longer working range, but the amount ofchange in muscle fiber length that occurs as a jointrotates depends on the muscle moment arm, the“mechanical advantage” that a muscle has at a par-ticular joint. This idea can be explained by compar-

FIGURE 8. Schematic drawing of two muscles with different PC-SAs but identical mass. (A) Comparison of isometric length–tension properties. (B) Comparison of isotonic force–velocityproperties. The effect of increased PCSA with identical fiberlength is to shift the absolute length–tension and force–velocitycurves to higher values but with retention of the same range andintrinsic shape.


ing the situation in which a simulated “muscle” isattached to a joint using two different moment armsbut muscles with identical fiber lengths. In the casewhere the moment arm is greater, the muscle fiberswill change length much more for a given change injoint angle compared with the same change in jointangle in the system with shorter fibers. As a result,the active ROM for the muscle-joint system with thelarger moment arm will be smaller compared withthe system with the larger moment arm in spite ofthe fact that their muscular properties are identical.Therefore, the design of a muscle when consideredin isolation may or may not be directly related to theactual function of a muscle once placed in the skel-etal system.

It is now important to qualify the previous state-ment about muscle design and architecture. Musclesthat are designed for speed based on their long fi-bers may not actually produce large velocities if theyare placed in the skeleton with a very large momentarm. The increased moment arm results in a verylarge joint moment, so that the muscle would behighly suited for torque production but at low angu-lar velocities. Similarly, a muscle that appears to bedesigned for force production due to the largePCSA, if placed in position with a very small momentarm, may actually produce high joint excursions orangular velocities. Differences between muscle-jointsystems thus require complete analysis of both jointand muscular properties (see below).

ISOKINETIC DYNAMOMETERS USED INPHYSICAL ASSESSMENT

It is necessary to characterize human performanceobjectively, not only to evaluate patient progress butalso to ascertain the efficacy of clinical treatment.Using objective criteria such as maximum joint mo-ment, shape of the moment–angle relationship, andROM, it is possible to evaluate the efficacy of manysurgical and rehabilitative procedures or the pro-gression of neuromuscular diseases. One of the mostcommonly used tools for musculoskeletal assessmentis the isokinetic dynamometer. If joint velocity is setto zero, an isometric joint moment versus joint–angle relationship is generated which, as describedabove, represents the interaction between muscleand joint properties. It is clearly not appropriate toascribe any portion of a moment versus joint anglecurve to either muscle or joint properties alone.Such moment–joint angle curves are also often gen-erated at constant angular velocities, hence the term,“isokinetic.” It is assumed that constant angular ve-locity corresponds to constant muscle velocity. How-ever, this is a poor assumption for a number of rea-

sons: (1) joint moment arm is not always constant,and therefore linear changes in joint angle do notcorrespond to linear changes in either muscle ormuscle fiber length; (2) muscle activation is neitherinstantaneous, constant, nor maximal, all of whichare required to generate the classic force–velocityrelationship; and (3) the compliance of the tendonpermits muscle length changes that are not propor-tional to joint angle changes. Although these pointshave been known for some time, nonconstantmuscle velocity was directly demonstrated experi-mentally by Fukunaga and colleagues,19 who usedultrasonography to measure fascicle length changesduring isokinetic tasks. The authors found that, forthe reasons stated above, the only parameter con-stant about isokinetic motion was joint angular ve-locity. Muscle fiber velocity, tendon velocity, andmuscle velocity all changed in a highly nonsteadystate manner.

Influence of PCSA and Fiber Length on IsokineticTorque. It is reasonable to assume that joint kine-matics are velocity independent and, thus, that varia-tions in the moment–angle curves as a function ofvelocity represent variations in muscle force. Themoment achieved during isokinetic contraction(concentric muscle contraction) must necessarily beless than that achieved during isometric contraction,due to the muscle’s force–velocity relationship. Also,muscle force generated during an isokinetic contrac-tion is a function of the muscle’s PCSA and fiberlength. The reason that isokinetic muscle force var-ies with PCSA is obvious because force is directlyproportional to PCSA. The reason that muscle forcevaries with fiber length is less obvious, althoughstraightforward. The isokinetic moment measure-ments are obtained while the joint is limited to aspecific isokinetic movement. This means that themuscle is also forced to maintain a certain (but notnecessarily constant) shortening velocity. Becauseshortening velocity is fixed (because excursion isfixed), the longer are the muscle fibers, the higher isthe muscle force that can be sustained during short-ening. This is because longer muscle fibers havemore sarcomeres in series and, with longer fibers,each sarcomere has a slower absolute contractionvelocity, allowing it to stay higher on its force–velocity curve (i.e., closer to Po). This idea is illus-trated in the force–velocity curve of Figure 9, whereforce is measured from two hypothetical musclesthat are shortening at the same velocity. Note that,for a given shortening velocity (vertical dotted line),the muscle with longer fibers maintains a higherforce compared with the muscle with shorter fibers.


This means that the muscle with longer fibers gen-erates the greater force at all velocities.

Earlier, it was stated that a muscle with longerfibers was designed for speed or excursion, whereasa muscle with a high PCSA was designed for forceproduction. It is now apparent that in the intact mus-culoskeletal system, this is not necessarily the case.

Theoretical Calculation of Optimal Architecture.The previous discussion highlights the general prin-ciple that force production in muscle is highly de-pendent on architecture, although not necessarily inways that are intuitively obvious. This is becausemuscle velocity achieved during motion has a very

large effect on force production. It is not effective to“design” a high-force muscle by giving it a hugePCSA and short Lf, if it is placed in a situation whereshortening velocity is sizable. This is because theshort fibers have a high sarcomere shortening veloc-ity and thus force depression. We investigated thetrade-off between PCSA and Lf in producing muscleforce for the mouse tibialis anterior muscle, whosearchitectural7 and kinematic properties wereknown.10 We considered the various ways in whichthis muscle could be “designed,” given the constraintof constant volume (or mass) within the shank. Themain result was that an optimal fiber length can beshown to be long enough to permit low sarcomerevelocity but short enough to permit high PCSA. Thelogic is as follows: If muscle fiber length of the tibialisanterior is increased while maintaining muscle mass,PCSA (and thus isometric muscle force) necessarilydecreases (Fig. 10A). This is a detriment to forceproduction. In contrast, as fiber length increases atconstant muscle mass, sarcomere velocity decreases,resulting in increased muscle force at this velocity(Fig. 10B). These two effects oppose one another,resulting in a theoretically “optimal” fiber length,representing the simultaneous satisfaction of bothcriteria. By calculating the tension expected duringmuscle shortening as a function of calculated fiberlength variation, the interaction between isometricforce production (resulting from PCSA) and dy-namic force production (sarcomere number and ve-locity) can be seen (Fig. 10C). The two opposingeffects in Figures 10A and 10B result in the existenceof an “optimal” fiber length, because at longer fiberlengths, PCSA continues to decrease hyperbolically,whereas sarcomere velocity decreases linearly. Thisdiminishing sarcomere velocity increases force as fi-ber length increases, but the concomitant reducedPCSA decreases force with increasing fiber length,and the latter effect outweighs the former. At shorterfiber lengths, the increased sarcomere shorteningvelocity overpowers the increased isometric force po-tential, again resulting in lower dynamic force pro-duction. For the mouse tibialis anterior, dynamictension was calculated to maximize at a fiber length(8.1 mm) that was very close to that observed bydirect measurement (7.6 mm).30 The tension pro-duced by the observed 7.6-mm fiber was still 99% ofmaximum. These data reinforce the idea that insome muscles that experience cyclic and stereotypicactivity, architecture is optimized to achieve maximalforce production under the precise conditions thatthe muscle experiences. In the previous study con-sidered,30 fiber length was optimized in the soleusand medial gastrocnemius muscles, as well.

FIGURE 9. Schematic drawing of two muscles with different fiberlengths but identical PCSAs. (A) Comparison of isometric length–tension properties. (B) Comparison of isotonic force–velocityproperties. The effect of increased fiber length is to increase theabsolute range of the length–tension curve and absolute velocityof the force–velocity curve but with retention of the same peakforce and intrinsic shape. Dotted vertical line demonstrates that,for an equivalent absolute velocity, the muscle with longer fibersgenerates a greater force.


ARCHITECTURAL DESIGN OFSYNERGISTIC MUSCLES

Given the specific design of individual muscles, itmight be expected that muscle placed in similar lo-cations within the body might have similar architec-tures. This is a difficult statement to support and, infact, a number of examples in the literature demon-strate that two muscles of divergent architectures areplaced in synergistic locations.

Human Wrist Extensors. One fairly detailed ex-ample of divergent architecture was seen in the longand short extensors of the human wrist: the extensorcarpi radialis longus (ECRL) and extensor carpi ra-dialis brevis (ECRB). The ECRL muscle is theshorter muscle of the two but contains longer musclefibers. The Lf/Lm ratio is relatively high (∼0.8) andPCSA relatively low (∼1.5 cm2), leading to the asser-tion that the ECRL is a muscle designed for highexcursion or velocity. The ECRB, on the other hand,is a longer muscle with shorter fibers and, therefore,a much lower Lf/Lm ratio (∼0.4) but a higher PCSAcompared with the ECRL (∼2.7 cm2). These datasuggest that, from a muscle point of view, theECRB is designed preferentially for high force pro-duction. Using rigid body kinematics, we quantifiedthe moment arms of both muscles acting in wristextension.42 We wondered whether architectural dif-ferences between muscles might actually be compen-sated for by changes in wrist moment arm (r). Wealso measured the sarcomere length change in eachmuscle during wrist rotation in patients undergoingsurgery for tennis elbow.37 Sarcomere length changeduring joint rotation (i.e., dSL/dv) is a direct reflec-tion of the relative ratio of the Lf:r ratio in a muscu-loskeletal system.68 The interesting result of thisstudy was that the skeletal kinematics actually accen-tuated the architectural differences betweenmuscles. The extensor moment arm of the ECRB wasmuch greater throughout the range of motion com-pared with that of the ECRL. Thus, muscle fiberlength change with joint rotation was expected to begreater in the ECRB compared with the ECRL. Thiswas confirmed by the intraoperative sarcomerelength measurements that showed an average dSL/dv for the ECRB of 9.1 nm/degree, whereas that ofthe ECRL was approximately half this value, or only4.7 nm/degree. This led to a mechanical model ofthe wrist extensors (Fig. 11) explaining that the largedSL/dv of the ECRB was due to the large momentarm and short fibers, whereas the small dSL/dv ofthe ECRL was due to the small moment arm andlong fibers. Using the architectural properties of thetwo muscles,34 along with nominal values for the

FIGURE 10. (A) Theoretical relationship between muscle fiberlength and isometric force for a constant muscle mass. As fiberlength increases, PCSA and thus Po decreases nonlinearly. (B)Theoretical relationship between muscle fiber length and dy-namic muscle force at a constant shortening velocity. Dynamicforce increases due to decreasing sarcomere velocity. (C) Pre-dicted force calculated to be generated by a tibialis anterior asfiber length is altered. Optimal fiber length results from a trade-offbetween decreasing sarcomere shortening velocity (which in-creases force) and decreasing muscle isometric force capacity asfiber length increases. (Data from Lieber.31)


Vmax of mammalian muscle,11 the force–velocity re-lationship for each muscle was also calculated. Thecontraction velocity at which the ECRL becomesstronger was calculated to be ∼80 mm/s which, onthe basis of the two muscle moment arms, corre-sponded to an angular velocity of ∼240°/s. This re-sults in a design in which the ECRB is stronger iso-metrically but, as angular velocity increases, theECRL becomes the stronger muscle.

Differential muscle strength that is velocity-dependent may provide insight into the diversity ofarchitectural design among muscles. The ECRB andECRL muscles as a synergistic group can generate amaximum tetanic tension (based on the sum of thePCSAs) of 25.6 kg and can produce a maximum an-gular velocity (based on the length of the ECRL fi-bers) of approximately 2800°/s. In order for a singlemuscle to generate that much force while maintain-ing the same Vmax, the fiber length would have to be76 mm and the cross-sectional area would have to be4.2 cm2. Using eq. 1 to calculate muscle physiologi-cal cross-sectional area,59 this single muscle wouldweigh 33.7 g, which is over 30% greater than the sumof the two muscle masses.34 Having two muscles assynergists thus accomplishes the same task at the ve-locity extremes but with a lower mass than a compa-rable, single, “supermuscle.” The design lesson isthat it is more efficient from a mass point of view tohave more highly specialized muscles than a single“supermuscle” that can accomplish the task of mul-tiple muscles.

Rabbit Dorsiflexors. A similar disparity of architec-tural properties is also seen in the rabbit dorsiflexormuscles: the tibialis anterior and extensor digitorumlongus. The tibialis anterior has relatively long fibers

(∼4.0 cm) and a small PCSA (0.6 cm2), whereas theextensor digitorum longus has shorter fibers (∼1.5cm) and a higher PCSA (1.4 cm2).32 Interestingly,the former muscle crosses only the ankle joint,whereas the latter crosses the ankle, tarsal, metatar-sophalangeal, and phalangeal joints. One might ex-pect that, in general, muscles that cross multiplejoints would have longer muscle fibers in order toprovide the functional range required when numer-ous joints move simultaneously. In the case of thetibialis anterior and extensor digitorum longus ofthe rabbit, this is not the arrangement. The rationalefor such diversity is not clear. As with the humanECRB/ECRL muscles, were a single “supermuscle”to be designed with the force-generating capacity ofthe sum of the two muscles and the overall excursionof the long tibialis anterior, the total mass would besignificantly greater by 45%. Additionally, if synergis-tic motion of the ankle and toes were to routinelyoccur during the gait cycle, then short fibers of theextensor digitorum longus would not be a limitationand might simply represent a design that permitspacking of a large number of fibers into a small vol-ume, to generate high force.

Cat Triceps Surae. Architectural disparity in thecat triceps surae between the medial gastrocnemiusand soleus muscles may have a more obvious func-tional role.60,63 Cat soleus muscle fibers are fairlylong: the soleus is composed of 100% slow musclefibers and the muscle crosses only the ankle joint. Incontrast, the medial gastrocnemius is composed ofabout 80% fast fibers, has relatively short fibers, andcrosses both the ankle and knee joints. Based onjoint kinematics measured during walking, it hasbeen shown that the cat knee and ankle joints movein opposite directions during normal gait, whichmeans that the medial gastrocnemius muscle lengthchange during locomotion is minimized. Based onthe known mechanical properties of muscle, highshortening velocities are unfavorable because theyresult in low muscle force, even when the muscle isfully activated. Therefore, the gastrocnemius canproduce high muscle force, due to its large numberof fibers packed at a relatively high pennation angle,even though the fibers are relatively short.

The soleus muscle is fairly active even during sta-tionary standing in the cat and is nearly maximallyactivated when the cat is walking relatively slowly.63

Because the soleus is uniarticular, moderate lengthchanges and thus high shortening velocities are ob-served during locomotion, which can compromisemuscle force. Long soleus muscle fibers “oppose”this tendency. That the soleus is composed of only

FIGURE 11. Schematic diagram of the interrelationship betweenfiber length and moment arm for the ECRB and ECRL torquemotors. The ECRB (bold print, thick lines) with its shorter fibersand longer moment arm changes sarcomere length about 2.5times as much as the ECRL, with its longer fibers and smallermoment arm. (Adapted from Lieber et al.37)


slow fibers makes the force depression due to short-ening even more pronounced, as the muscle has arelatively low Vmax. The triceps surae in the cat thusrepresents a system in which fiber type and musclearchitecture appear to have developed in a comple-mentary manner.

MUSCULOSKELETAL BALANCE

The previous discussion focuses primarily on torqueproduction of a single musculoskeletal system. How-ever, much of normal movement is predicated oneffective and coordinated interaction between op-posing muscle groups. Detailed muscle and skeletaldata on agonist–antagonist pairs are largely lacking.However, based on the relatively complete dataset regarding wrist muscle34 and joint42 propertiesthat we have collected, a few general trends haveemerged. Based on the intraoperative sarcomerelengths measured,37,38 wrist extensors were pre-dicted to operate primarily on the plateau and de-scending limb of their sarcomere length–tensioncurve, with all muscles generating maximal force infull extension. Only the ECRB was predicted to op-erate at sarcomere lengths corresponding to lessthan 80% Po in the normal range of motion. Wristflexors (flexor carpi ulnaris and flexor carpi radialis)were predicted to operate predominantly on theshallow and steep ascending limbs of their length–tension curve, with both major flexors generatingmaximal force in full wrist extension. In full flexion,it was possible for wrist flexors to generate forces thatwere less than 50% Po only.

Such a design presents interesting implicationsfor the design of the wrist as a torque motor. Bothflexor and extensor muscle groups generate maxi-mum force with the wrist fully extended. As the wristmoves from flexion to extension, maximum extensorforce increases due to extensor shortening up thedescending limb of the length–tension curve andmaximum flexor force increases due to flexorlengthening up the ascending limb of the length–tension curve. This effect is superimposed upon anincreasing extensor moment arm as the extensormuscles elevate off of the wrist under the extensorretinaculum and a decreasing flexor moment armas the flexors juxtapose the wrist from the flexorretinaculum. Combining muscle and joint effects,extensor muscle force is amplified by an increasingextensor moment arm and flexor muscle force isattenuated by a decreasing flexor moment arm. In-terestingly, because the flexors as muscles developsignificantly greater force than do the extensors(due to their larger PCSA), the net result is a nearlyconstant ratio of flexor to extensor torque over the

wrist range of motion. In fact, although at the mus-cular level, the flexors are considerably strongerthan the extensors,5,34 when including the wrist ki-nematics, extensor moment actually slightly exceedsflexor moment. Additionally, wrist resistance to an-gular perturbation increases as the wrist is moved tofull extension, because both flexor and extensor mo-ments increase in a similar fashion. This is anotherway of saying that the wrist is most mechanicallystable in full extension, which is probably the reasonthat individuals who are asked to perform a powergrip operation do so with the wrist extended. Itshould be noted that wrist torque balance is achievedat the expense of maximum moment generation atthe wrist, and, therefore, we conclude that this mus-culoskeletal system is not designed simply to operatenear maximum force, as is often assumed in muscu-loskeletal models.

PLASTICITY OF MUSCLE ARCHITECTURE

The fact that architecture varies between muscleswithin a species and is extremely consistent within aspecies implies that regulation of serial sarcomerenumber in muscles is ongoing. It is easy to demon-strate that muscle serial sarcomere number (andthus fiber length) varies rapidly with such treatmentsas immobilization,66 tenotomy,1 moment arm alter-ation,29 and even eccentric exercise.45 However, al-though muscle adaptation to these conditions is con-sistent and well described, the mechanical andcellular factors to which a muscle responds underthese conditions are not yet fully understood.

In muscles that are used in stereotypical fashion,muscle architecture and joint kinematics seem to in-teract such that muscles produce near-maximalpower during normal use, as described above. Forexample, the complex architecture of fish muscle54

and the spatial segregation of fish red (slow) andwhite (fast) muscle53 enable fish to maintain near-optimal sarcomere lengths and near-optimal powerproduction at velocities normally seen during loco-motion.53 The very large frog semimembranosusmuscle, which powers hopping, appears to have theappropriate architecture such that during the hop,the muscle generates near-maximal power at near-optimal sarcomere lengths.43,44 Optimization of thisrelationship is not trivial. Because the moment armtransforms the linear muscle motion into angularjoint rotation, the moment arm influences both themuscle length range and shortening velocity re-quired to produce a given joint motion. Thus, theinteraction between muscle architectural propertiesand joint moment arms is critical for producing thecorrect joint kinematics during locomotion.


Immobilization Studies of Architecture Alteration.One of the most simple and straightforward demon-strations of architectural change occurs secondary tolimb immobilization. In a classic study that hasgreatly influenced modern thinking on this subject,Williams and Goldspink66 immobilized the ankles ofyoung (8-week-old) mice. Ankles were held in a fixedposition for 3 weeks by means of a plaster cast, andthe properties of the monoarticular soleus musclewas measured at the end of the immobilization pe-riod. The result was straightforward: regardless ofthe position of immobilization, the muscle changedits length–tension properties so that maximum iso-metric tension was observed at the muscle lengthcorresponding to the angle of immobilization. Thesedata could indicate that muscles “adapt” to chroniclength changes by resetting sarcomere length to theoptimum or simply that muscles adapt to reset themuscle to a specific sarcomere length. Architecturalanalysis revealed that this change in resting lengthwas accompanied by a change in serial sarcomerenumber within the soleus fibers (Fig. 12). For ex-ample, whereas normal soleus muscles had about2200 serial sarcomeres, when the ankle was immobi-lized in dorsiflexion (thus stretching the muscle),sarcomere number increased to almost 2600, andwhen the ankle was immobilized in plantarflexion(thus shortening the muscle), sarcomere numberdecreased to ∼1800. Thus, the mouse soleus musclehad adjusted to its new mechanical environment byadding or deleting the appropriate number of sar-comeres such that optimal muscle length coincidedwith the length at which the muscle was immobi-lized.

Retinaculum Release Studies of Architecture Alter-ation. The studies of Williams and Goldspink66

have been extended in concept by many physiolo-gists and surgeons to indicate that muscles simplyadjust serial sarcomere number to optimize forceproduction. However, there are a number of reasonsnot to accept this generalization. First, it was dem-onstrated that although the soleus muscle may showsuch adaptations, not all muscles adapt in the samemanner.61 Thus, the basic nature of the adaptationwas similar in the medial gastrocnemius as well as thetibialis anterior, but the magnitude of the responsewas greatly attenuated (Fig. 13). This study suggeststhat different muscles have different degrees of re-sponsiveness to sarcomere number change. Thisshould not be surprising, because different musclegroups demonstrate large differences in their ten-dency to either atrophy or hypertrophy in responseto an altered mechanical environment.14,35,57

A difficulty in interpreting results from immobi-lization studies is that it is impossible to identifythe particular signal to which sarcomere numberadapts.24 Sarcomere number may change within amuscle fiber so that optimal sarcomere length is de-termined by resting muscle length, by resting muscletension, by an extreme of muscle length, or even bythe length of most frequent or maximal muscle ac-tivation. Chronic length change models cannot dis-tinguish between these possibilities. In an effort toextend this body of knowledge, length changes havebeen imposed upon muscles that are not immobi-

FIGURE 12. Sarcomere number in mouse soleus muscles of8-week-old mice subjected to immobilization. Serial sarcomerenumber was determined after 3 weeks of immobilization with themuscle in the control, shortened, or lengthened position. (Datafrom Williams and Goldspink.66)

FIGURE 13. Fiber length of rat muscles immobilized in the neu-tral (filled bars), lengthened (open bars), or shortened (hatchedbars) position. Note that muscles respond to immobilization in thesame direction although with different magnitudes. (Data fromSpector et al.61)


lized, by transecting the ankle retinaculum and per-mitting the muscle an increased active excursionduring normal movement. This procedure was im-posed upon rapidly growing rabbits, and it wasshown that muscle excursion was dramatically in-creased, as were whole-muscle length and fiberlength, all in proportion to the increased excur-sion.12 Unfortunately, the interpretation of these ex-periments is confounded by the rapid growth rate ofthese animals, which itself increases fiber length.Retinaculum release was performed in a separatestud, and the mechanical environment of the musclewas studied in vivo in much greater detail.29 Inter-estingly, although the serial sarcomere number in-crease was also documented, the force generated bythe released tibialis anterior as well as its PCSA ac-tually decreased. These data may represent simulta-neous adaptation of serial and parallel sarcomerenumber to complex mechanical stimuli. Neither thetorque produced by the released tibialis anterior northe ankle flexor moment arm were reported, but it isinteresting to speculate that the tibialis anterior de-creased its PCSA proportional to the increased mo-ment arm to keep dorsiflexion torque constant.

In light of the functional importance of fiberlength and the general lack of understanding of me-chanical and biological factors that regulate fiberlength, we developed a model in which the demandsplaced upon the muscle were altered so that sarco-mere number could adapt to either length or excur-sion in a way that could be differentiated.9 The anklepretibial flexor retinaculum was surgically tran-sected, and the tibialis anterior muscle was permit-ted to move away from the center of rotation of theankle joint, increasing the tibialis anterior momentarm and decreasing muscle–tendon unit (MTU)length as the foot dorsiflexes. The mechanical effecton the muscle can be viewed as a chronic shorteningas well as an increased excursion and velocity. Priorto retinaculum transection, the moment arm wasrelatively constant throughout the range of motion,as would be expected, because the tendon of thetibialis anterior passes beneath the flexor retinacu-lum. After transection, there was a notable peak dur-ing dorsiflexion due to the altered geometry allow-ing the tibialis anterior to “bowstring” in operatedanimals. The major effect of transection was to in-crease the moment arm at joint angles of less than100°. Based on moment arm measurements, averageMTU excursion increased from 1.4 mm in controlanimals to 1.9 mm in the transected groups over thefull range of motion (50–150°). Despite the signifi-cant increase in maximum muscle length and signifi-cant decrease in minimum muscle length, both of

which might be expected to lead to increased sarco-mere number, sarcomere number decreased signifi-cantly over the 2-week experimental period, from3200 ± 200 in control animals to 2900 ± 200 in mice2 weeks after transection. Similar results were ob-tained after 2 weeks of swim training following reti-naculum transection (2950 ± 170; n = 4), indicatingthat lack of tibialis anterior activation did not inducethe results.

Video analysis of stepping in normal and reti-naculum-transected mice revealed that 2-week tran-sected mice preferentially dorsiflexed at slightlyslower and more variable rates. Unoperated micedorsiflexed fairly consistently, at 2200 ± 300°/s,which was significantly faster than 2-week retinacu-lum-transected mice (1900 ± 400°/s). Despite this14% decrease in angular velocity after transection,the 30% increase in moment arm and significantreductions in dorsiflexion and plantarflexion musclelengths resulted in a calculated increase of approxi-mately 20% in the velocity of tibialis anterior short-ening during dorsiflexion, from 2.7 to 3.3 Lo s−1. Theeffect of altered sarcomere number is also reflectedin the decreased power production calculated at thepreferred gait speed. Whereas control mice were cal-culated to generate 92 ± 2% of peak tibialis anteriormuscle power, 2-week transected mice were calcu-lated to produce only 74 ± 5% of their peak musclepower in tibialis anterior. These data suggest that notall architectural adaptations that occur in a complexmechanical environment result in “reoptimization”of muscle properties. In the mouse retinaculum-release study, the only parameter that appeared toreturn to the initial condition was the minimum sar-comere length. Taken at face value, these data sug-gest that the muscle adapted in response to the un-derlying chronic change in length rather than thealtered excursion or velocity. Obviously, the detailsof the mechanical factors regulating sarcomere num-ber in muscle remain to be elucidated.

Architectural Change after Eccentric Contraction.One of the most extreme mechanical conditions un-der which a muscle may operate is forced lengthen-ing of an activated muscle, i.e., eccentric contrac-tion, when the muscle is not usually used in thatmanner. Chronic eccentric training of a muscle re-sults in a number of changes, most of which arebeyond the scope of this review. However, based ona theory that suggests that sarcomere length operat-ing range should remain on the ascending limb andplateau of the length–tension curve, Morgan46 pre-dicted that a muscle subjected to eccentric trainingshould add sarcomeres in series. This would have the


effect of shifting the muscle operating range toshorter lengths, perhaps even those on the ascend-ing limb. This hypothesis was tested in an experi-ment in which rats were trained to either rundownhill or uphill or remain sedentary. The down-hill-running protocol, due to the extreme knee flex-ion angles reached, resulted in eccentric training ofthe vastus intermedius muscle. The uphill-trainingprotocol controlled for the effect of exercise itself.45

The authors demonstrated an increase in serial sar-comere number from about 3250 in sedentary con-trols to about 3500 in decline runners. This sup-ported their underlying assumption that musclesadapt such that they will not be stretched on thedescending limb of the length–tension curve. Al-though these data explain the behavior of the ratvastus intermedius, some muscles appear to be de-signed to operate on the descending limb of thelength–tension curve. For example, based on intra-operative sarcomere length measurements, the hu-man ECRB operates almost exclusively on the de-scending limb of the length–tension curve duringwrist joint rotation.38 Thus, it is difficult to define astereotypical sarcomere length operating range formuscles. In a recent survey of 83 separate studiescovering a total of 51 different muscles in eight dif-ferent species, the average minimum sarcomerelength was 82 ± 17% of optimal length (Lo) and anaverage maximum length of 119 ± 21% Lo

8; 90% ofmaximum sarcomere lengths fell within 92–167%Lo. The average sarcomere length of 100 ± 14% Lo

across the entire review suggests that Lo ± 14% Lo

would make a reasonable first approximation for a“generic” muscle operating range in the absence ofother data.

In summary, this review has presented basic defi-nitions of skeletal muscle architecture as well as asummary of the architectural properties of humanupper and lower limb muscles. The varying architec-tural design of human and other mammalianmuscles was used to illustrate the fact that musclescan be “designed” to perform fairly specific func-tions. Finally, muscle architecture can change withaltered muscle use or in the face of a new mechani-cal environment. A detailed understanding of themechanical and biological factors that regulate suchadaptation will undoubtedly be the source of futurestudies.

REFERENCES

1. Baker JH, Hall CE. Changes in sarcomere length followingtenotomy in the rat. Muscle Nerve 1980;3:413–416.

2. Bodine SC, Roy RR, Meadows DA, Zernicke RF, Sacks RD,

Fournier M, Edgerton VR. Architectural, histochemical, andcontractile characteristics of a unique biarticular muscle: thecat semitendinosus. J Neurophysiol 1982;48:192–201.

3. Boyes JH. Tendon transfers for radial palsy. Bull Hosp JointDis 1960;21:97–105.

4. Brand PW. Tendon transfers for median and ulnar nerve pa-ralysis. Orthop Clin North Am 1970;1:447–454.

5. Brand PW, Beach RB, Thompson DE. Relative tension andpotential excursion of muscles in the forearm and hand. JHand Surg (Am) 1981;3A:209–219.

6. Burke RE. Motor units: anatomy, physiology, and functionalorganization. In: Peachey LD, editor. Handbook of physiol-ogy. Bethesda: American Physiology Society; 1981. p 345–422.

7. Burkholder TJ, Fingado B, Baron S, Lieber RL. Relationshipbetween muscle fiber types and sizes and muscle architecturalproperties in the mouse hindlimb. J Morphol 1994;220:1–14.

8. Burkholder TJ, Lieber RL. Brief review: sarcomere length op-erating range of muscles during movement. Submitted.

9. Burkholder TJ, Lieber RL. Sarcomere number adaptation af-ter retinaculum release in adult mice. J Exp Biol 1998;201:309–316.

10. Burkholder TJ, Lieber RL. Sarcomere number regulation bylength and velocity. Med Sci Sports Exerc 1996;28:S167.

11. Close RI. Dynamic properties of mammalian skeletal muscles.Physiol Rev 1972;52:129–197.

12. Crawford GNC. An experimental study of muscle growth inthe rabbit. J Bone Joint Surg Am 1954;36A:294–303.

13. Eccles JC, O’Connor WJ. Responses which nerve impulsesevoke in mammalian striated muscles. J Physiol (Lond) 1939;97:44–102.

14. Edgerton VR, Barnard RJ, Peter JB, Maier A, Simpson DR.Properties of immobilized hind-limb muscles of the galagosenegalensis. Exp Neurol 1975;46:115–131.

15. Edman KAP, Reggiani C, Te Kronnie G. Differences in maxi-mum velocity of shortening along single muscle fibres of thefrog. J Physiol (Lond) 1985;365:147–163.

16. English AW. An electromyographic analysis of compartmentsin cat lateral gastrocnemius muscle during unrestrained loco-motion. J Neurophysiol 1984;52:114–125.

17. Friden J, Lieber RL. Suitability of the posterior deltoid-to-triceps tendon transfer based on muscle architectural prop-erties. J Hand Surg (in press).

18. Friederich JA, Brand RA. Muscle fiber architecture in thehuman lower limb. J Biomech 1990;23:91–95.

19. Fukunaga T, Ichinose Y, Ito M, Kawakami Y, Fukashiro S.Determination of fascicle length and pennation in a contract-ing human muscle in vivo. J Appl Physiol 1997;82:354–358.

20. Gans C, Bock WJ. The functional significance of muscle ar-chitecture: a theoretical analysis. Adv Anat Embryol Cell Biol1965;38:115–142.

21. Gans C, De Vries F. Functional bases of fiber length andangulation in muscle. J Morphol 1987;192:63–85.

22. Goldner JL. Tendon transfers for irreparable peripheralnerve injuries of the upper extremity. Orthop Clin North Am1974;5:343–375.

23. Henneman E, Somjen G, Carpenter DO. Functional signifi-cance of cell size in spinal motorneurons. J Neurophysiol1965;28:560–580.

24. Herring SW, Grimm AF, Grimm BR. Regulation of sarcomerenumber in skeletal muscle: a comparison of hypotheses.Muscle Nerve 1984;7:161–173.

25. Huijing PA, Baan GC. Stimulation level-dependent length-force and architectural characteristics of rat gastrocnemiusmuscle. J Electromyogr Kinesiol 1992;2:112–120.

26. Jacobson MD, Raab R, Fazeli BM, Abrams RA, Botte MJ,Lieber RL. Architectural design of the human intrinsic handmuscles. J Hand Surg (Am) 1992;17A:804–809.

27. Kardel T. Willis and Steno on muscles: rediscovery of a 17th-century biological theory. J Hist Neurosci 1996;5:100–107.

28. Kawakami Y, Ichinose Y, Fukunaga T. Architectural and func-


tional features of human triceps surae muscles during con-traction. J Appl Physiol 1998;85:398–404.

29. Koh TJ, Herzog W. Excursion is important in regulating sar-comere number in the growing rabbit tibialis anterior. J Phys-iol (Lond) 1998;508:267–280.

30. Lieber RL. Muscle fiber length and moment arm coordina-tion during dorsi- and plantarflexion in the mouse hindlimb.Acta Anat 1997;159:84–89.

31. Lieber RL. Skeletal muscle structure and function: implica-tions for physical therapy and sports medicine. Baltimore:Williams & Wilkins; 1992. 303 pp.

32. Lieber RL, Blevins FT. Skeletal muscle architecture of therabbit hindlimb: functional implications of muscle design. JMorphol 1989;199:93–101.

33. Lieber RL, Brown CC. Quantitative method for comparisonof skeletal muscle architectural properties. J Biomech 1992;25:557–560.

34. Lieber RL, Fazeli BM, Botte MJ. Architecture of selected wristflexor and extensor muscles. J Hand Surg (Am) 1990;15A:244–250.

35. Lieber RL, Friden JO, Hargens AR, Danzig LA. Gershuni DH.Differential response of the dog quadriceps muscle to exter-nal skeletal fixation of the knee. Muscle Nerve 1988;11:193–201.

36. Lieber RL, Jacobson MD, Fazeli BM, Abrams RA, Botte MJ.Architecture of selected muscles of the arm and forearm:anatomy and implications for tendon transfer. J Hand Surg(Am) 1992;17A:787–798.

37. Lieber RL, Ljung B-O, Friden J. Intraoperative sarcomeremeasurements reveal differential musculoskeletal design oflong and short wrist extensors. J Exp Biol 1997;200:19–25.

38. Lieber RL, Loren GJ, Friden J. In vivo measurement of hu-man wrist extensor muscle sarcomere length changes. J Neu-rophysiol 1994;71:874–881.

39. Lieber RL, Ponten E, Friden J. Sarcomere length changesafter flexor carpi ulnaris-to-extensor digitorum communistendon transfer. J Hand Surg (Am) 1996;21A:612–618.

40. Loeb GE, Gans C. Electromyography for experimentalists.Chicago: University of Chicago Press; 1986. 371 p.

41. Loeb GE, Pratt CA, Chanaud CM, Richmond FJR. Distribu-tion and innervation of short, interdigitated muscle fibers inparallel-fibered muscles of the cat hindlimb. J Morphol 1987;191:1–15.

42. Loren GJ, Shoemaker SD, Burkholder TJ, Jacobson MD,Friden J, Lieber RL. Influences of human wrist motor designon joint torque. J Biomech 1996;29:331–342.

43. Lutz GJ, Rome LC. Muscle function during jumping in frogs.I. Sarcomere length change, emg pattern, and jumping per-formance. Am J Physiol 1996;271:C563–C570.

44. Lutz GJ, Rome LC. Muscle function during jumping in frogs.II. Mechanical properties of muscle: Implications for systemdesign. Am J Physiol 1996;271:C571–C578.

45. Lynn R, Morgan DL. Decline running produces more sarco-meres in rat vastus intermedius muscle fibers than does in-cline running. J Appl Physiol 1994;77:1439–1444.

46. Morgan DL. New insights into the behavior of muscle duringactive lengthening. Biophys J 1990;57:209–221.

47. Ounjian M, Roy RR, Eldred E, Garfinkel A, Payne JR, Arm-strong A, Toga AW, Edgerton VR. Physiological and develop-mental implications of motor unit anatomy. J Neurobiol 1991;22:547–559.

48. Petit J, Filippi GM, Emonet-Denand C, Hunt CC, Laporte Y.Changes in muscle stiffness produced by motor units of dif-ferent types in peroneus longus muscles of cat. J Neurophysiol1990;63:190–197.

49. Petit J, Filippi GM, Gioux C, Hunt CC, Laporte Y. Effects oftetanic contraction of motor units of similar type on the initialstiffness to ramp stretch of the cat peroneus longus muscle. JNeurophysiol 1990;64:1724–1731.

50. Powell PL, Roy RR, Kanim P, Bello M, Edgerton VR. Predict-ability of skeletal muscle tension from architectural determi-

nations in guinea pig hindlimbs. J Appl Physiol 1984;57:1715–1721.

51. Riordan DC. Tendon transfers for median, ulnar, or radialnerve palsy. J Bone Joint Surg Br 1968;50B:441–449.

52. Rome LC, Choi IH, Lutz G, Sosnicki A. The influence oftemperature on muscle function in the fast swimming scup. I.Shortening velocity and muscle recruitment during swim-ming. J Exp Biol 1992;163:259–279.

53. Rome LC, Funke RP, Alexander RM, Lutz G, Aldridge H,Scott F, Freadman M. Why animals have different muscle fi-ber types. Nature 1988;335:824–827.

54. Rome LC, Sosnicki AA. Myofilament overlap in swimmingcarp ii. Sarcomere length changes during swimming. Am JPhysiol 1991;163:281–295.

55. Roy RR, Bello MA, Powell PL, Simpson DR. Architecturaldesign and fiber type distribution of the major elbow flexorsand extensors of the monkey (cynomolgus). J Morphol 1984;171:285–293.

56. Roy RR, Bodine-Fowler SC, Kim J, Haque N, de Leon D,Rudolph W, Edgerton VR. Architectural and fiber type distri-bution properties of selected rhesus leg muscles: feasibility ofmultiple independent biopsies. Acta Anat 1991;140:350–356.

57. Roy RR, Medows ID, Baldwin KM, Edgerton VR. Functionalsignificance of compensatory overloaded rat fast muscle. JAppl Physiol 1982;52:473–478.

58. Roy RR, Powell PL, Kanim P Simpson DR. Architectural andhistochemical analysis of the semitendinosus muscle in mice,rats, guinea pigs, and rabbits. J Morphol 1984;181:155–160.

59. Sacks RD, Roy RR. Architecture of the hindlimb muscles ofcats: functional significance. J Morphol 1982;173:185–195.

60. Spector SA, Gardiner PF, Zernicke RF, Roy RR, Edgerton VR.Muscle architecture and force–velocity characteristics of thecat soleus and medial gastrocnemius: implications for motorcontrol. J Neurophysiol 1980;44:951–960.

61. Spector SA, Simard CP, Fournier M, Sternlicht E, EdgertonVR. Architectural alterations of rat hindlimbs skeletal musclesimmobilized at different lengths. Exp Neurol 1982;76:94–110.

62. Trotter JA, Purslow PP. Functional morphology of the endo-mysium in series fibered muscles. J Morphol 1992;212:109–122.

63. Walmsley B, Hodgson JA, Burke RE. Forces produced by me-dial gastrocnemius and soleus muscles during locomotion infreely moving cats. J Neurophysiol 1978;41:1203–1216.

64. Weeks OI, English AW. Compartmentalization of the cat lat-eral gastrocnemius motor nucleus. J Comp Neurol 1985;235:255–267.

65. Wickiewicz TL, Roy RR, Powell PL, Edgerton VR. Muscle ar-chitecture of the human lower limb. Clin Orthop 1983;179:275–283.

66. Williams P, Goldspink G. Changes in sarcomere length andphysiological properties in immobilized muscle. J Anat 1978;127:459–468.

67. Woittiez RD, Baan GC, Huijing PA, Rozendal RH. Functionalcharacteristics of the calf muscles of the rat. J Morphol 1985;184:375–387.

68. Zajac FE. How musculotendon architecture and joint geom-etry affect the capacity of muscle to move and exert force onobjects: a review with application to arm and forearm tendontransfer design. J Hand Surg (Am) 1992;17A:799–804.

69. Zuurbier CJ, Huijing PA. Changes in geometry of activelyshortening unipennate rat gastrocnemius muscle. J Morphol1993;218:167–180.

70. Zuurbier CJ, Huijing PA. Influence of muscle geometry onshortening speed of fibre, aponeurosis and muscle. J Biomech1992;25:1017–1026.


Documents

INVITED REVIEW - Freee.guigon.free.fr/rsc/article/LieberFriden00.pdf · INVITED REVIEW ABSTRACT: Skeletal muscle architecture is the structural property of ... “the arrangement