Exercise Science, LLC

Personal Training and Rehabilitation, New Orleans, LA.


Specificity of Speed in Exercise... A Vague Concept

Ryan A. Hall

Human Performance and Health Promotion

University of New Orleans


There has been tremendous debate on how strength training programs for athletes should be designed. Some feel that strength training should imitate the skill used in performance and that performing fast strength training movements will translate to quickness on the playing field. A review of the principles of physics, motor learning, and classical biology show that this thinking is flawed in many ways. Furthermore, recent findings indicate that the tools used to determine the relationship between strength and speed of movement are invalid. Strength exercise should be performed in accordance with muscle and joint function. A study comparing the percentage of strength increases using two different training protocols with different speeds of movement may suffice to further examine the relationship between strength and speed of movement.

Specificity of Speed in Exercise... A Vague Concept

Physics affects everything humans do. Especially in relation to exercise, physics plays a vital role in determining the effects it will have on the human body. An understanding of how physics interacts with human movement is essential in the testing and application of proper exercise methods. Gravity, inertia, velocity, acceleration, friction, momentum, work, power, and torque are all aspects of physics that must be considered in proper exercise application and measurement. Exercise researchers have been interested in how these factors affect the human body for some time now, mostly concerning speed of movement. What are the effects of speed, velocity, and acceleration on the body? And, what practical application does an understanding of these concepts have in relation to exercise? Research in this area has been equivocal, especially regarding how these concepts apply to exercise performance.

Do fast movements in the weight room make athletes fast on the football field? Some report that in order to be fast, one should train fast. Such concepts are the basis that isokinetic philosophy is built upon (Moffroid & Whipple, 1969, 1970). Although, others state that strength exercise should be controlled and in accordance with muscle and joint function (Pollock & Wilmore, 1991; Hutchins, 1992). There remains much controversy over this topic.

Earlier research seemed to support the notion of specificity of speed in strength training. (Moffroid Whipple, 1970; Cole et al., 1981) However, more recent literature has pointed out many design flaws in those studies, as well as serious problems and limitations in the measurement tools used in those studies (references). The purpose of this paper is to provide a review of the literature concerning muscle testing as it relates to strength and speed of movement, as well as examine how several aspects of physics apply to speed of human movement in strength exercise.

Review of Testing Procedures

In order to fully examine the concept of specificity of speed of movement as it relates to strength training, it is important to examine the measurement tools. All of the studies examining the relationships between speed of movement and strength used isokinetic equipment. Critical reviews throughout the 1980’s (Beimborn & Morrissey, 1987; Mayhew & Rothstein, 1985; Rothstein, Lamb, & Mayhew, 1987; Winter, Wells, & Orr, 1981) have all emphasized the lack of scientific evidence supporting isokinetic philosophy, as well as highlighted a number of technical problems with all isokinetic testing procedures: calibration, patient stabilization, axis placement, moment arm, gravity, damping, limited range, interval scaling, body weight normalization, and the use of reciprocal contractions. These reviews prompted many questions concerning the reliability and validity of isokinetic testing equipment.

Reliability is the degree of consistency that a measuring instrument or procedure demonstrates. There is conflicting evidence in the literature concerning the reliability of isokinetic measures. Early reliability studies all suffered from small sample sizes, inadequate reporting of data, inadequate statistics, or a combination of all of these (Newton & Waddell, 1993). One possible source of variability in test results, mentioned earlier, is axial alignment. Grabiner, Jeziorowski, and Divekar (1990) showed that changes in axial alignment could produce up to 20% variations in peak torque of isokinetic measures. Delitto, Rose, Crandell, and Strube (1991), in there study of isokinetic measures on a LIDO machine, found that torso flexion-extension ratios were unreliable with test-retest errors approaching 50% of mean scores. However, others (Portfield, Mostardi, & King, 1987) have shown that Cybex isokinetic testing gave readings to within 1% when subjects were retested on different days, although no detailed statistics were given. Other researchers (Smith, Mayer, Gretchel, & Becker, 1985) have shown no evidence for the reliability of isokinetic testing at speeds greater than 120° per second. This is in agreement with other current research that has shown isokinetic testing equipment to be reliable only at speeds slower than 120° per second (Newton & Waddell, 1993). It is important to note that all of the studies concerning strength and velocity testing on isokinetic equipment included speeds above 120° per second. Some of the studies tested exclusively at speeds above 120° per second, at which isokinetic measures have been shown to be unreliable.

Validity is the quality of an instrument that enables it to measure what it is supposed to measure. Moreover, it is possible for a test to be reliable, but still not be valid. This has shown to be the case, in some instances, with isokinetic testing equipment. There is no direct evidence on the relationship between isokinetic measures and actual muscle strength or functional capacity in realistic activities, only indirect evidence. Bemben, Grump, and Massey (1988) found that torque may be over estimated by up to 30% when the Cybex dynamometer is used for isokinetic assessment at higher velocities. Another fundamental limitation of isokinetic measurements is that the machine only records torque if the subject is able to move at the preset test velocity. A certain amount of muscle strength is required to move the trunk or limbs themselves, and a zero recording does not reflect true absence of muscle strength. This amount of torque is unmeasured, and thus omitted from all data recordings, leaving the recorded isokinetic torque unproportional to actual muscle strength.

There are other problems associated with dynamic testing on isokinetic equipment. Strength is usually defined as the maximum level of torque that is produced by the force of muscular contraction. Isokinetic equipment claims to measure torque produced by muscular contraction. However, there are several factors of non-muscular torque that can affect the results of any dynamic test of muscular strength. These factors must be accounted for, by either removing or factoring the measurements into the results, in order for the test to be accurate (Jones, 1993; Fulton, 1993).

Furthermore, in order for a test of muscular strength to be meaningful, the joint being tested must be totally isolated. Without total isolation, it is impossible to determine the actual source of muscular torque being measured (Jones, 1993; Graves et al., 1994; Graves et al., 1990a; Graves et al., 1990b; Pollock, Leggett, Graves, Jones, Fulton, & Cirulli, 1989). For example, the hip extensors (mainly the gluteals and hamstrings) work together to move the pelvis toward extension of the back. Although these muscles may be important, they are not responsible for extension of the lumbar spine. In order to accurately measure torque produced by the lumbar extensors, pelvic movement must be blocked. In test of lumbar extension, any movement of the pelvis will be confused with movement of the lumbar spine, thus producing inaccurate test results. Isokinetic testing equipment does not allow for total isolation of the joints on any of the testing equipment. Upon witnessing an isokinetic test, it can be clearly seen that other structures of the body are moving in addition to the joint being tested.

Gravity is another very obvious component of non-muscular torque. The weight of the subject’s body part or parts being tested must be counterweighted or the non-muscular torque produced will bias the results (Jones, 1993; Fulton, 1993; Graves et al., 1990a) (see Fig 1). This is an absolute requirement for meaningful test of strength and one that is ignored by isokinetic testing devices.

Stored energy is another source of non-muscular torque that must be considered in a true test of strength. Movement in any direction away from an anatomically neutral, relaxed position will result in compression of soft tissues on one side of the joint being tested and stretching of soft tissue on the other side of the joint being tested. This compression and stretching of soft tissue will produce stored energy. This will result in very high levels of non-muscular torque that will tend to move the involved body part or parts back towards the neutral position. This, however, can not be removed. But it can be measured and factored into the test results. This factor is also ignored by isokinetic testing devices.

Other sources of non-muscular torque that are ignored in all dynamic isokinetic test of strength are friction and impact forces (Jones, 1993; Fulton 1993; Graves et al., 1990a). Impact forces are an unavoidable, unwanted, and often dangerous aspects of dynamic isokinetic testing. They tend to overstate true levels of muscular strength and can be harmful when testing not only injured, but healthy joints as well. Moreover, Isokinetic testing equipment is based on friction. Isokinetic tests involve muscular movement against friction. However, this is yet another source of non-muscular torque. Not only does apparatus (or machine) friction interfere with the test results, but intramuscular friction will bias the results also. Intramuscular friction is the friction produced from muscular contraction. As a muscle contracts to move an involved body part through a range of motion, actin and myosin filaments slide against each other. Any sliding between two surfaces causes friction, even inside muscle tissue. However, the solution to these two problems is simple. By using a static or isometric testing procedure, the effects of friction are removed (Jones, 1993; Fulton, 1993). If there is no movement, there is no effect of friction. Impact forces are also avoided during a properly conducted isometric test. However, the nature of any isokinetic testing procedure is dynamic. Therefore, these two components of non-muscular torque are ignored by any isokinetic test.

In conclusion, there is a lack of evidence to support that isokinetic tests measure what the manufacturers claim they measure. There is no theoretical basis and no valid experimental evidence to support claims that isokinetic measures at different speeds measure different characteristics of muscle physiology, such as strength at slow speeds or power at high speeds (Mayhew & Rothstein, 1985). Jerome, Hunter, Gordon, & McKay (1991) found isokinetic torque, work, and power were all highly correlated (0.75 - 0.98). On factor analysis they extracted a single variable that accounted for 89% of the total variance and concluded that different isokinetic measures all measured essentially the same variable. They called this variable the muscle performance index and found it to be independent of test velocity. There is also considerable evidence suggesting that isokinetic testing equipment produces meaningless test results by ignoring factors of non-muscular torque, such as involvement of other muscular structures, gravity, stored energy, impact forces, and friction. These factors are principles of basic mechanical physics and can not be ignored when assessing true muscular strength.

Specificity of Movement Speed

Early research regarding specificity of movement speed and muscle strength started in the 1970’s. All of such research was performed on isokinetic testing equipment. One of the landmark studies on this topic, Specificity of Speed of Exercise (Moffroid & Whipple, 1970), is a good example. This study is one of the foundations of isokinetic research and philosophy. The researchers investigated the effects of slow and fast resistance exercise on strength gains at various velocities. Three groups (group I - slow speed training, group II - fast speed training, and the control group, which received no training) were tested on two occasions (before and after training) on a Cybex isokinetic dynamometer, which supposedly measured peak torque output at various velocities. The authors concluded that slow speed training produced increases in muscular strength only at slow speeds, and that fast speed training produced increases in muscular strength at both slow and fast speeds.

However, upon further examination of their data, it can be seen quite clearly that the authors stated the negative correlation of their own published data, rather than the positive correlation. What the data really shows is that slow speed training produced gains at both tested slow and fast speeds and that little increases in strength were made at tested slow or fast speeds with fast speed training.

Cole et al. (1981) also demonstrated results that were similar to the conclusions made by Moffroid and Whipple. Although, a further analysis of this study is needed to fully explore the mechanisms involved. The researchers trained three groups at various velocities on an isokinetic knee extension exercise: (1) slow group - 60° per second, (2) fast group - 300° per second, and (3) mixed group - both 60° per second and 300° per second. Each of the groups were tested before and after training at three velocities: 60° per second, 180° per second, and 300° per second. The results were that the slow group significantly improved peak isokinetic torque only at 60° per second. The fast group significantly improved peak torque at 300° per second and 180° per second. And, the mixed group improved significantly at only the training speeds, 60° per second and 300° per second, but not 180° per second.

There are many problems associated with these studies. First of all, there is significant evidence, reviewed earlier, that shows isokinetic testing to be unreliable at speeds greater than 120° per second. Both studies used training and testing velocities greater than 120° per second. Secondly, dynamic isokinetic testing overlooks the five factors of non-muscular torque. These factors can not be ignored if test of muscular strength are to be accurate. Thirdly, a knee extension exercise machine is approximately a 110-120° range of motion exercise, depending on the manufacturer. Even at speeds of 60° per second, which is considered slow to the researchers, the entire movement will be completed in less two seconds. When compared to movement speeds recommended by others (Hutchins, 1992) that involve a movement speed of ten seconds for the concentric part of the exercise, 60° speed of lifting certainly does not seem slow. At speeds of 180° per second, the entire movement will be completed in less than one second. And, at speeds of 300° per second, the movement would be so rapid that the entire range of motion of the knee extension exercise would be completed in a little over 1/3 of a second (.367 seconds). Few subjects can discriminate between movement speeds less than one second. There is evidence to suggest that movement at such fast speeds is injurious, due to impact forces (Boocock, Garbutt, Linge, Reilly, & Troup 1990).

Also, neither of these studies considered motor learning as a factor in increasing peak torque in the test itself before reporting baseline measures. A learning effect has been clearly documented in isokinetic testing (Newton & Waddell, 1993; Mawdsley & Knapik 1982). Both studies recommended using at least the second test session for baseline measures. Smidt et al. (1985), in an early reliability study noted 13 - 21% increases on repeat testing after one week with no training. Grabiner, Jeziorowski, and Divekar (1990) retested subjects five minutes apart on a Biodex isokinetic torso flexion/extension machine and showed an increase in measured peak torque ranging from 0-28%. Delitto et al. (1991) found a 5 - 15% increase in isokinetic peak torque on a LIDO torso flexion/extension machine in three tests during three weeks with no training. Hence, a single baseline testing session may seriously underestimate isokinetic torque. Few studies allow for this motor learning effect. It is very probable that in these studies a significant increase in measured torque on isokinetic testing equipment was due to motor learning and not gains in muscular strength.

Most of the evidence suggests that the specificity effect is due to improvements in neurological ability or skill improvements, not muscular adaptations. Knapik and Ramos (1980) found a high relationship between isometric and slow speed isokinetic test. However, correlations decreased as testing speeds became more widely separated. They explained the effect by stating that as isokinetic velocities move farther apart, the motor task become more dissimilar, requiring different patterns of neural recruitment and coordination. Behm and Sale (1993) suggest that neural or skill adaptations play a major role in the specificity effect of movement speed. They based their conclusions upon research on ballistic activities (a very high speed contraction). Activity of the motor neurons involved in ballistic contractions has been shown even during the pre-movement silent period.

Thorstenson (1977) showed evidence against velocity specificity. In this study, the researcher compared two subjects who were either progressively resistance trained with maximal or near maximal weights or detrained over a five month period. After the first eight weeks of training, peak torque showed an improvement of 25% over all tested speeds except for the highest speed, which only exhibited an increase of 10%. After five months of training, there were further increases in peak torque at all speeds, with the largest increase at the highest speed. Although the sample size (n=2) is insufficient to produce statistical inferences to a population, this study argues against velocity specificity, illustrating the possible increases in torque at all velocities with intense exercise. This is in agreement with others that show maximal fiber recruitment of both slow and fast twitch fibers as exercise intensity increases.

What Physics Tells Us...

Cole et al. (1981) suggest that patients in rehabilitation should perform fast movements, such as knee extensions at 300° per second, because training at fast speeds enhances strength at both fast and slow speeds. They also state that high tension contractions occur with the use of slow training. Therefore, faster training is again recommended for rehabilitation patients because of their low tolerance for high tension contractions. From the standpoint of mechanical physics, this thinking is flawed in many ways.

Injury to muscular structures and other connective soft tissues is caused by unnecessary and abrupt increases in force (American Academy of Orthopaedic Surgeons, 1991; Heyward, 1991; McArdle, Katch, &Katch, 1991). Force is a product of two factors: mass and acceleration. Hence, the equation: F = m * a (Giancoli, 1986). Acceleration equals velocity divided by time (a=v/t). It is not the actual velocity that is the culprit for causing dangerous increases in acceleration, and therefore dangerous increases in force. Rather, it is how quickly the change in velocity occurs. Change in velocity is the definition of acceleration (Giancoli 1986). The longer it takes the velocity to change, the less acceleration, and therefore the less force is produced on the body. The quicker the velocity changes, the more acceleration, and therefore the more force is produced on the body. Since fast speed weight training involves rapid changes in velocity, from start to stop of each repetition, often with excessive masses, this provides the most likely scenario for injury. There is sufficient evidence to support the fact that high impact forces lead to injury (Boocock et al, 1990; American Academy of Orthopaedic Surgeons, 1991; Pollock & Wilmore, 1990).

What Motor Learning Research Tells Us...

Specificity, as a concept, was first founded exclusively in the field of motor learning research. The notion that strength training performed at fast speeds will transfer to quickness on the playing field in athletic events, or that strength training movements should simulate the action or skill of the sport being played is contradictory to motor learning research. Skills are highly specific. Activities used for practice may either help, harm, or have no effect on athletic performance. Basically, there are three types of transfer: positive, negative, and indifferent. Positive transfer occurs when the activities of practice and actual performance are identical. Negative transfer occurs when the activities of practice are almost, but not identical to, the activities of performance. Indifferent transfer occurs when the activities of practice and performance are totally unrelated (Schmidt, 1991).

In order for positive transfer to occur, the activity performed in practice should be exactly the same as the performance event. It is a mistake to make practice almost the same as the performance event. Such slight changes will result in negative transfer and cause slight, barely noticeable disturbances in a good athlete’s performance. However, it may confuse his neuromuscular patterns enough to make the difference between a good performance and a great performance. It is important to realize that skill conditioning is not the same as physical conditioning.


In conclusion, the notion of specificity of speed in strength exercise is a vague concept. Principles of physics, motor learning, biomechanics, and physiology are in conflict with the notion of specificity of speed in exercise. The theory that in order to be fast, one must train fast, is incorrect. This thinking not only overlooks several principles of mechanical physics, but also unnecessarily raises the potential risk of injury. Muscle fibers are recruited on the basis of need, not speed. High intensity exercise has been shown to recruit more of both slow twitch and fast twitch fibers, not one or the other. Strength training should be performed in accordance with full range muscle and joint function. To load a skill in order to become stronger at that activity is a mistake. This notion is contrary to the specificity principle of motor learning and will lead to negative transfer. Moreover, since sports skills normally incorporate tremendous amounts of acceleration, loading the skill causes unnecessary increases in force, thereby increasing the potential risk of injury. Furthermore, all research to date on the specificity theory of speed in exercise was performed using dynamic isokinetic testing, which has been shown to be both unreliable, at most testing speeds, and invalid. Isokinetic testing does not measure true muscle strength. These testing procedures do not take into consideration several aspects of non-muscular torque that will bias any test of dynamic strength.


American Academy of Orthopaedic Surgeons. (1991). Athletic training and sports medicine. Rosemont, IL: American Academy of Orthopaedic Surgeons.

Behm, D. & Sale, D. (1993). Velocity specificity of resistance training. Sports Medicine, 15(6), 374-388.
Bemben, M., Grump, K., & Massey, B. (1988). Assessment of technical accuracy of the Cybex II isokinetic dynamometer and analog recording system. Journal of Orthopaedic and Sports Physical Therapy, 10, 12-17.

Biemborn, D., & Morrissey, M. (1987). A review of the literature related to trunk muscle performance. Spine, 13, 655-660.

Boocock, M., Garbutt, G., Linge, K., Reilly, T., & Troup, J. (1990). Changes in stature following drop jumping and post exercise gravity inversion. Medicine and Science in Sports and Exercise, 22(3), 385-390.

Coyle, E., Feiring, D., Rotkis, T., Cote, R., Roby, F., Lee, W., & Wilmore, J. (1981). Specificity of power improvement through and fast isokinetic training. Journal of Applied Physiology, 51(6), 1437-1442.

Delitto, A., Rose, J., Crandell, C., & Strube, M. (1991). Reliability of trunk muscle performance. Spine, 16, 321-324.

Fulton, M. N. (1993). Spinal rehabilitation (part 1): measuring true functional ability in clinical practice. Ocala, Fl: MedX Corporation.

Giancoli, D. C. (1986). The ideas of physics. Orlando, FL: Harcourt Brace Jovanovich, Inc.

Grabiner, M., Jeziorowski, J., & Divekar, A. (1990). Isokinetic measurements of trunk extension and flexion performance collected with the Biodex clinical data station. Journal of Orthopaedic and Sports Physical Therapy, 11, 590-598.

Graves, J., Pollock, M., Carpenter, D., Leggett, S., Jones, A., MacMillan, M., & Fulton, M. (1990a). Quantitative assessment of full range of motion isometric lumbar extension strength. Spine, 15(4), 289-294.

Graves, J., Pollock, M., Foster, D., Leggett, S., Carpenter, D., Rosemaria, V., & Jones, A. (1990b). Effect of training frequency and specificity on lumbar extension strength. Spine, 15(6), 504-509.

Graves, J., Webb, D., Pollock, M., Matkozich, J., Leggett, S., Carpenter, D., Foster, D., & Cirulli, J. (1994). Pelvic stabilization during resistance training: Its effect on the development of lumbar extension strength. Archives of Physical Medicine and Rehabilitation, 75, 210-215.

Heyward, V. H. (1991). Advanced fitness assessment and exercise prescription. Champaign, IL: Human Kinetics Books.

Hutchins, K. (1992). Super slow: the ultimate exercise protocol. Casselberry, FL: Media Support.

Jerome, J., Hunter, K., Gordon, P., & McKay, N. (1991). A new robust index for measuring isokinetic trunk flexion and extension - Outcome from a regional study. Spine, 16, 804-808.

Jones, A. (1993). The lumbar spine, the cervical spine, and the knee. Ocala, Fl: MedX Corporation.
Knapik, J., & Ramos, M. (1980). Isometric and isokinetic torque relationships in the human body. Archives of Physical Medicine and Rehabilitation, 61, 64-67.

Mawdsley, R. & Knapik, J. (1982). Comparison of isokinetic measures with test repetitions. Physical Therapy, 62, 169-172.

Mayhew, T. & Rothstein, J. 1985. Measurement of muscle performance with instruments. Measurement in Physical Therapy. New York, Churchill Livingston, 57-99.

McArdle, W. D., Katch, F. I., & Katch, V. L. (1991). Exercise physiology: energy, nutrition, and human performance. Malvern, PA: Lea & Febiger.

Moffroid, M. & Whipple, R. (1970). Specificity of speed in exercise. Physical Therapy, 50(12), 1692-1700.

Moffroid, M., Whipple, R., Hofkosh, J., Lowman, E., & Thistle, H. (1969). A study of isokinetic exercise. Physical Therapy, 49(7), 735-746.

Newton, M., & Waddell, G. (1993). Trunk strength testing with iso-machines. Spine, 18(7), 801-811.

Pollock, M., Leggett, S., Graves, J., Jones, A., Fulton, M., & Cirulli, J. (1989). Effect of resistance training on lumbar extension strength. American Journal of Sports Medicine, 17(5), 624-629.

Pollock, M. L., & Wilmore, J. H. (1990). Exercise in health and disease: evaluation and prescription for prevention and rehabilitation. Philadelphia, PA: W. B. Saunders Company.

Portfield, J. A., Mostardi, R. A., King, S. (1987). Simulated lift testing using computerized isokinetics. Spine, 12, 683-687.

Rothstein J., Lamb, R., & Mayhew, T. (1987). Clinical use of isokinetic measurements: critical issues. Physical Therapy, 67, 1840-1843.

Schmidt, R. A. (1991). Motor learning and performance: from principles to practice. Champaign, IL: Human Kinetics Books.

Smith, S., Mayer, T., Gatchel, R., & Becker, T. (1985). Quantification of trunk lumbar function Part 1: Isometric and multispeed isokinetic trunk strength measures in sagittal and axial planes in normal subjects. Spine, 8, 757-764.

Thorstensson, A. (1977). Observations on strength training and detraining. Acta Physiologica Scandinavia, 100, 491-493.

Winter, D., Wells, R., & Orr, G. (1981). Errors in the use of isokinetic dynamometers. European Journal of Applied Physiology, 46, 397-408.