Secondary Logo

Journal Logo

Original Research

Anthropometric and Metabolic Determinants of 6,000-m Rowing Ergometer Performance in Internationally Competitive Rowers

Mikulic, Pavle

Author Information
Journal of Strength and Conditioning Research: September 2009 - Volume 23 - Issue 6 - p 1851-1857
doi: 10.1519/JSC.0b013e3181b3dc7e
  • Free



Rowing athletes are tested to evaluate their individual sport-specific performance and to detect changes in performance capacity. For that purpose, rowing ergometers, which are widely considered valuable tools in testing, are commonly used by rowers, who vary in terms of age, sex, quality level, and classification. Measuring the shortest time needed to cover a particular distance is perhaps the easiest approach to testing and evaluating exercise capacity. In addition to the frequently used “all-out” ergometer tests over a 2,000-m distance, which corresponds to the actual distance used for Olympic rowing events, the test over a 6,000-m distance is widely conducted (7,11,15,19).

The duration of a 2,000-m rowing ergometer test closely corresponds to the duration of 2,000-m “on-water” races (especially for bigger boats, such as quads, fours, and eights), thereby simulating the metabolic demands of the actual “on-water” performance. Performance over 2,000-m on a rowing ergometer is dependent on the functional capacity of both aerobic and anaerobic energy pathways, with the relative energy derived from anaerobic metabolism being 21 to 30% (25). The 6,000-m rowing ergometer test is assumed to primarily estimate a rower's aerobic fitness (19). That is why this test is generally conducted during the first part of the preparatory period, which usually starts in October, and is aimed at building up the aerobic base through aerobic extensive endurance training, which occupies approximately 90% of the total training time during that preparation period (18).

The importance of the rowing technique is less pronounced when rowing on an ergometer in comparison with on-water rowing, which allows for the better isolation of the anthropometric and metabolic variables and their importance in determining performance time. The use of laboratory-based prediction models allows coaches to predict on-water performance and to identify potentially talented rowers (22). Performance-predictive parameters for 2,000-m rowing ergometer performance have been established in many studies (3,10,12,13,16,21,22,30) examining performance standard, sex, and competitive classifications. To the author's knowledge, 6,000-m rowing ergometer performance remains unexamined. The results of such a study could yield valuable information that subsequently could be used for general rowing performance modeling, rowing talent identification, and selection. The aim of this study, therefore, is to examine the anthropometric and metabolic determinants of performance during 6,000-m of rowing on an ergometer in male heavyweight rowers.

The hypothesis was established that a combination of anthropometric and metabolic variables would be a better predictor of 6,000-m rowing ergometer performance than either category of variables or individual variable alone. This hypothesis is based on the following argument: Because, in previous studies (3,10,12,13,16,21,22,30), certain anthropometric and metabolic variables were found to be related to 2,000-m rowing performance, it might be expected that a combination of variables from both sets (anthropometric and metabolic) would lead to a larger proportion of explained variance, therefore providing the more accurate prediction models for rowing ergometer performance over a 6,000-m distance.


Approach to the Problem

Because the determinants of 2,000-m rowing ergometer performance have been extensively explored, and with 6,000-m rowing ergometer performance remaining unexamined, this study was designed to identify the anthropometric and metabolic determinants of 6,000-m rowing ergometer performance in male heavyweight rowers. For this purpose, the rowers participating in the present study underwent anthropometric measurements and, using the incremental maximal test on a rowing ergometer, measurements of their metabolic capacities. The measurements were performed in a controlled laboratory environment. Anthropometric and metabolic variables were chosen based on the characterization of various components of body build (28) and on the measurements used in related studies of male and female rowing athletes (1,5,10,12,21).

The rowers' 6,000-m rowing ergometer performance, the dependent variable in the present study, was evaluated based on the results of the 2007 Croatian Indoor Rowing Championship, which took place within 15 days of the laboratory testing. Pearson correlation coefficients and stepwise multiple linear regression were used to determine the relationship between the measured anthropometric and metabolic variables and 6,000-m rowing ergometer performance. Statistical power was calculated according to the method developed by Hopkins (9), which is based on the statistical power of 0.8 and the level of significance of 0.05; for a detection of a correlation coefficient of 0.5 the study should comprise approximately 25 subjects. The reliability data for 6,000-m ergometer performance are not currently available; however, reliability data for 2,000-m ergometer performance indicate high reliability and stability over time in trained rowers (23,27). Accordingly, it could be assumed that 6,000-m performance on the same measurement apparatus (Concept II, model C rowing ergometer, Morrisville, Vermont) would also produce satisfactory reliability and stability data.


The sample comprised 25 current or former Croatian national rowing or sculling champions and members of Croatia's national team (mean ± SD: age 22.2 ± 4.8 years, rowing experience 8.8 ± 4.6 years, stature 1.91 ± 0.05 m, body mass 91.7 ± 5.9 kg). All the subjects were highly trained, internationally experienced rowers competing for a spot on the national team for the 2007 rowing season. Each subject participating in this study had a history of competing as a member of Croatia's national team in 1 or more of the following events: FISA World Junior Championships, FISA World Under-23 Championships, FISA World Rowing Championships, or the Olympic Games. As part of the testing process, each subject was asked to give his written informed consent following an explanation of the nature and purpose of the experiment and of the risks associated with participation. This explanation was in compliance with the Declaration of Helsinki. All experimental procedures were approved by the Ethics Committee of the School of Kinesiology, University of Zagreb. The laboratory measurements and the 6,000-m time trial were performed in January, in the midst of the preparatory period for the rowers on the northern hemisphere.

Experimental Procedures

In accordance with the recommendations of the The International Society for the Advancement of Kinanthropometry (17), the anthropometry was performed as follows: stature, body mass, arm span, arm relaxed girth, gluteal thigh girth, chest girth, and 6 skinfolds (triceps, subscapular, supraspinale, abdominal, front thigh, and medial calf). The percentage of body fat was estimated using the Carter equation (2) used in the study of body composition in the Montreal Olympic Games Anthropological Project. Lean body mass was calculated by subtracting estimated body fat from total body mass. In addition, forced vital capacity was measured using the Quark b2 spirometry system (Cosmed, Rome, Italy). After the anthropometric measurements were taken, the rowers warmed up according to their usual habits and completed an incremental maximal test on the Concept II model C rowing ergometer. The main purpose of incremental maximal tests is to determine individual maximal exercise capacity. The test started with a 3-minute rowing session at 150 W and was subsequently intensified by 25-W increments per minute. When a rower was no longer able to maintain the required power for 10 seconds, he was instructed to perform a 30-second “all out” effort to ensure that he achieved maximal levels of aerobic capacity.

The Quark b2 breath-by-breath gas exchange system (Cosmed), equipped with Quark b2 6.0 PC software support, was used to collect and analyze expired air. Heart rate was monitored using the short-range Polar radio telemetry system (Polar Electro OY, Kempele, Finland). Cardiorespiratory parameters were calculated automatically and printed every 30 seconds. The highest values were calculated as the arithmetic mean of the 2 consecutive highest 30-second values. Ventilatory threshold (VT) was determined noninvasively by combining 3 common methods for the determination of gas exchange thresholds (6): (a) the ventilatory equivalent of O2 and CO2 (VE·VO2−1 and VE·VCO2−1, respectively); (b) excess CO2; and (c) the V-slope method. The combination of these 3 methods of gas exchange threshold detection has been shown to improve the accuracy and reliability of VT identification (6). Two independent experienced investigators detected VT and, in the event of a disagreement, the opinion of a third investigator was sought, and the threshold was determined by consensus.

The variables obtained during the continuous progressive rowing ergometer test were as follows: maximal oxygen uptake (O2max; in L·min−1 and mL·kg−1·min−1), power output at maximal oxygen uptake (PVO2max, in W), maximal ventilation (Emax, in L·min−1), ventilatory threshold (VT, in % of O2max), oxygen uptake at ventilatory threshold (O2VT, in L·min−1 and mL·kg−1·min−1), and power output at ventilatory threshold (PVT, in W).

The 6,000-m rowing ergometer performance data were based on the results of the 2007 Croatian Indoor Rowing Championship that took place within 15 days following the laboratory testing. The rowers used “warm-up ergometers” to prepare for the race according to their usual habits and were then assigned to a “race ergometer” on which up to 10 rowers representing individual competition categories raced simultaneously. All rowing ergometers were the same make and type (Concept II, model C), and all “race ergometers” were linked to a central computer so that the competitors were able to view monitors that displayed their times and their relative positions in the race.

Statistical Analyses

The SPSS 11.5 for Windows statistical package (Chicago, Illinois) was used to process and report the data, which are reported as mean ± SD. A significance level of p < 0.05 was selected. Pearson correlation coefficients (r) were used to determine the strength of association of each of the independent variables and their relationship to the 6,000-m rowing ergometer time. Variables that correlated highly with performance were selected for the development of the regression models using stepwise multiple linear regression analysis. The significance (probability) of F value was used as the criteria for variable entry to the model and variable removal from the model. A variable entry criterion was set at p < 0.05, whereas a variable removal criterion was set at p < 0.10. The adjusted R2, as opposed to the sample R2, was used to assess the proportion of variance that could be explained by the independent variables. According to Russell et al. (22), the adjusted R2 considers both the number of predictor variables and the sample size. This results in the R2 being approximately corrected for the upward bias of the sample R2, which subsequently provides a more accurate estimate of goodness-of-fit of the prediction model.

Limitations of the Study

One limitation of this study must be acknowledged and addressed: The performance measurements were not obtained in a controlled laboratory environment. Indeed, the author believed that real performance (i.e., the athletes giving their best effort) is only evident in a competitive environment that includes direct competitors and encouragement from coaches and other team members. In addition, the selected methodology used for out-of-laboratory performance data collection has been used before when establishing performance-predictive parameters for 2,000-m rowing ergometer performance, namely in the studies conducted by Ingham et al. (10) and Riechman et al. (21).


Descriptive Statistics and Correlations

The results of the anthropometric and metabolic variables and their correlations with the 6,000-m rowing ergometer performance values have been calculated (Tables 1 and 2). It should be noted that performance was significantly correlated with the following anthropometric variables: body mass, lean body mass, forced vital capacity, arm span, relaxed arm girth, chest girth, and gluteal thigh girth. In the metabolic category of variables being observed, performance was correlated with maximal oxygen uptake (L·min−1), power output at maximal oxygen uptake, oxygen uptake at ventilatory threshold (L·min−1), and power output at ventilatory threshold. To permit a better understanding of the strength of the relationship among variables, scatterplot graphs with regression lines representing the variables that best explain performance variance (lean body mass, power output at maximal oxygen uptake, and power output at ventilatory threshold) are presented (Figure 1).

Table 1:
Anthropometric characteristics and 6,000-m rowing ergometer time and correlations with 6,000-m rowing ergometer time for internationally competitive heavyweight rowers.
Table 2:
Responses to the incremental maximal test on the rowing ergometer and correlations with 6,000-m rowing ergometer time for internationally competitive heavyweight rowers.
Figure 1:
Relationship among 6,000-m rowing ergometer performance time and (a) lean body mass, (b) power output at ventilatory threshold, and (c) power output at maximal oxygen uptake.

Regression Models

Seven anthropometric measures (body mass, lean body mass, forced vital capacity, arm span, relaxed arm girth, chest girth, and gluteal thigh girth) underwent a stepwise multiple regression analysis to predict the 6,000-m rowing ergometer time. Lean body mass was the first and only predictor to enter the model before the statistical limits (p < 0.05) were reached (Table 3). The formula used for the anthropometric prediction model (adjusted R2 = 0.575; standard error 22.8 seconds) is as follows:

Table 3:
Stepwise multiple linear regression models of 6,000-m rowing ergometer performance-individual and combined categories of variables.

Four metabolic variables [O2max (L·min−1), PVO2max, O2VT (L·min−1), and PVT] also underwent a stepwise multiple regression analysis to predict the 6,000-m rowing ergometer time, and power output at ventilatory threshold was shown to be the strongest and, ultimately, only true predictor in that metabolic sector (Table 3). The formula used for the metabolic prediction model (adjusted R2 = 0.530; standard error 24.7 seconds) is as follows:

Finally, a combination of variables (from anthropometric and metabolic categories) underwent a stepwise multiple regression analysis, which increased the predictive power of the model. The first variable to enter the formula was PVT, followed by FVC (Table 3). The formula used to establish the combined categories' best prediction model (adjusted R2 = 0.722; standard error 15.4 seconds) is as follows:


The present study has suggested that the strongest overall correlate of 6,000-m rowing ergometer performance is lean body mass (r = −0.767), followed by power output at ventilatory threshold (r = −0.743) and power output at maximal oxygen uptake (r = −0.732). Stepwise multiple linear regression procedures identified the equation containing power output at ventilatory threshold and forced vital capacity as the best performance-prediction model for 6,000-m rowing ergometer performance in internationally competitive heavyweight rowers (adjusted R2 = 0.722). The hypothesis that a combination of anthropometric and metabolic variables would be a better predictor of 6,000-m rowing ergometer performance than either category of variables or individual variable alone was supported.

The model comprising a combination of anthropometric and metabolic variables is followed by equations comprising anthropometric and metabolic variables separately, which produced adjusted R2 values of 0.575 for the anthropometric and 0.530 for the metabolic group of variables. Based on the results of the anthropometric equation, it can be concluded that 6,000-m rowing ergometer performance increases with body size. More specifically, large lean body mass contributes to a higher level of rowing performance. During rowing almost every muscle is used (25), and lean body mass frequently has been found to be a major predictor of rowing performance over 2,000 m (3,13,16). It is no coincidence that elite male and female rowers often are tall and have high lean body mass values (1,10, 21). Of note, this finding in elite rowers as the highly trained power-endurance athletes seems contrary to the inverse relationship between body size and endurance performance (as assessed using a cycle ergometer) obtained in untrained men (20).

A moderate correlation between the anthropometric equation and performance time may be explained by the relationship between morphology and rowing performance-that is, the performance of rowers is, in part, determined by their physical characteristics (26). Rowers who are tall and have high lean body mass values possess an advantage over their smaller and lighter peers because taller individuals are able to use a longer levering effect. Longer limbs that accompany the body size seem to be especially advantageous for performance with respect to the larger leverage and power output. In addition, larger rowers possess a larger cross-sectional area of muscle and a greater absolute metabolic capacity (5).

One of the surprising findings of this study is that maximal oxygen uptake (in L·min−1), which has consistently been found to be the strongest and often the only predictor of rowing ergometer 2,000-m performance (3,14,21,30), produced a significant but relatively low correlation over 6,000-m performance (r = −0.484) and was excluded from the prediction models involving both the metabolic and combined categories. Although surprising, this finding is in line with a previous suggestion that O2max may not correlate well with performance among a relatively homogeneous population of endurance athletes (30). In addition, perhaps a more precise way of saying that high O2max is decisive for rowing performance would be instead slightly above average O2max per unit body mass in subjects with high lean body mass, hence with high absolute O2max. Large O2max expressed in L·min−1 in rowers probably reflects larger body mass or, more specifically, large lean body mass.

Power output at ventilatory threshold proved to be the only performance-predictive parameter when metabolic variables were observed. According to Jensen (11), power depends on aerobic and anaerobic energy supplies balanced by efficiency or technique. Power at anaerobic threshold frequently has been found to be a major predictor of 2,000-m rowing ergometer performance. However, “threshold” has largely been determined using blood lactate responses to ergometry testing, rather than ventilatory (gas exchange) responses as was the case in the present study. In a review discussing the biology and medicine of rowing, Shephard (26) emphasized that rowing performance is closely related to the power output at the “anaerobic threshold.” Power output at ventilatory threshold and forced vital capacity, in the combined model, explained a total of 72.2% of performance variance. Vital capacity in rowers typically is great (26), with the lungs of rowers reflecting their large body dimensions. The highest reported value among world class rowers was found to be 9.1 L (24). It has been established that rowing training does not appear to influence total vital capacity once maximum body height has been reached (4). The combination of 2 performance predictors (i.e., power output at ventilatory threshold and forced vital capacity) requires relatively short testing time to be obtained, which should be considered an advantage of this prediction model. It is always the intention to collect the required information from the least possible number of tests and in the least possible time, so that rowing coaches and athletes do not have to spend too many hours in the laboratory at the expense of training (8).

Practical importance of the standard error of the estimate values that were obtained for regression models deserves to be briefly discussed. As shown in Table 3, the standard error of the estimate for the best multiple regression model is 15.4 seconds. If this value is transferred into practice and made applicable for real-life conditions, it is reasonable to expect that it is small enough to considerably differentiate between rowers' rankings in the majority of 6,000-m rowing ergometer competitions. For example, in the present study the difference between the first and the last ranked rower is 144.8 seconds, or more than 9 times the standard error of the estimate. Therefore, rowing coaches, athletes, or both can use the equation because it has a reasonably strong practical ability to predict the outcome of a 6,000-m race.

A comparison of the results of the present study with published prediction models (a rowing ergometer distance of 2,000 m) reveals that many researchers have found either O2max or PVO2max to be an important parameter in predicting 2,000-m rowing ergometer time (3,10,12,30). In addition, body mass and lean body mass are often described as important anthropometric predictors (3,12,22). For example, Cosgrove et al. (3) reported that lean body mass demonstrated the highest correlation (r = 0.85) with average velocity during a 2,000-m ergometer performance test, which is in line with the results of the present study. Jürimäe et al. (12) demonstrated identical findings, reporting lean body mass (r = −0.91) and PVO2max (r = −0.97) as the strongest correlates with performance over an ergometer distance of 2,000 m. Ingham et al. (10) reported PVO2max as the strongest correlate (r = −0.93) with rowing performance. In contrast, Russell et al. (22) reported that a combination of anthropometric variables (body height, body mass, and the sum of skinfold measurements) provided the best model for predicting ergometer performance.

The ability to predict performance is known to vary directly with the heterogeneity of the sample (29). In the present study, the sample consisted of male individuals reflecting a range of ages, competitive rowing abilities, and experience levels, varying from successful junior rowers who had recently embarked on their senior careers to Olympic medalists. Therefore, it was expected that the laboratory measurements would yield somewhat greater statistical power in measuring 6,000-m ergometer performance-predictive factors. However, it should be noted that the use of multiple regression procedures on a relatively small sample may reflect differences between individuals rather than the predictive value of laboratory assessments, although many studies designed to establish performance-prediction parameters in rowing have, to the best of the author's knowledge, used small-sample methodology, with the exception of the study conducted by Ingham et al. (10).

This study is, nevertheless, an attempt to contribute to the area of rowing performance modeling by providing data pertaining to internationally ranked male heavyweight rowers because information that specifically relates to performance predictors and correlates other than over 2,000-m rowing ergometer distance is generally absent from the available literature. The results of the present study must be interpreted with caution because the prediction equations were developed specifically for male heavyweight rowers. To use these predictors in a wider range of athletes such as with female rowers, lightweight rowers, or junior rowers, cross-validation in an independent sample of rowers of a particular category would be necessary. It also should be noted that the prediction equations are only as accurate as the tests used to measure the anthropometric and metabolic variables.

Practical Applications

Rowing ergometer performance over 6,000 m in internationally competitive heavyweight rowers is mainly determined by power output corresponding to the ventilatory threshold. The findings of the present study suggest that rowers striving to improve their 6,000-m rowing ergometer time should devote their training time to the improvement of lean body mass (which proved to be the strongest single correlate with 6,000-m ergometer performance) and to the improvement of power output corresponding to ventilatory threshold. These findings are of particular interest to coaches and rowers and are especially relevant to designing proper training sessions because both of the aforementioned determinants of performance are known to be highly trainable and controllable.


The author would like to thank the Croatian Rowing Federation for assistance and organizational help during the completion of this project. Also, gratitude is extended to Dr. Goran Markovic for critical comments and suggestions in the preparation of the manuscript.


1. Bourgois, J, Claessens, AL, Vrijens, J, Philippaerts, R, Van Renterghem, B, Thomis, M, Janssens, M, Loos, R, and Lefevre, J. Anthropometric characteristics of elite male junior rowers. Br J Sports Med 34: 213-216, 2000.
2. Carter, JEL. Body composition of Montreal Olympic athletes. In: Physical Structure of Olympic Athletes-Part I. Carter, JEL, ed. Basel: Medicine Sport Science, 1982. pp. 107-116.
3. Cosgrove, MJ, Wilson, J, Watt, D, and Grant, SF. The relationship between selected physiological variables of rowers and rowing performance as determined by a 2000m ergometer test. J Sports Sci 17: 845-852, 1999.
4. Danuser, HJ and Buhlmann, AA. Effect of regular training on total and vital capacity of the 17 to 25-year-old rowers. Schweiz Med Wochenschr 113: 454-458, 1983.
5. De Rose, EH, Crawford, SM, Kerr, DA, Ward, R, and Ross, WD. Physique characteristics of Pan American Games lightweight rowers. Int J Sports Med 10: 292-297, 1989.
6. Gaskill, SE, Ruby, BC, Walker, AJ, Sanchez, OA, Serfass, RC, and Leon, AS. Validity and reliability of combining three methods to determine ventilatory threshold. Med Sci Sports Exerc 33: 1841-1848, 2001.
7. Hagerman, FC. The physiology of competitive rowing. In: Exercise and Sport Science. Garrett, W Jr and Kirkendall, DT, eds. Philadelphia: Lippincott Williams & Wilkins, 2000. pp. 843-873.
8. Hahn, A, Bourdon, P, and Tanner, R. Protocols for the physiological assessment of rowers. In: Physiological Tests for Elite Athletes. Gore, CJ, ed. Champaign, IL: Human Kinetics, 2000. pp. 311-326.
9. Hopkins, WG. Estimating sample size for magnitude-based inferences. Sports Sci 10: 63-70, 2006. (Available at:
10. Ingham, SA, Whyte, GP, Jones, K, and Nevill, AM. Determinants of 2000 m rowing ergometer performance in elite rowers. Eur J Appl Physiol 88: 243-246, 2002.
11. Jensen, K. Test procedures for rowing. Lausanne: FISA, 1994.
12. Jürimäe, J, Mäestu, J, Jürimäe, T, and Pihl, E. Prediction of rowing performance on single sculls from metabolic and anthropometric variables. J Hum Mov Stud 38: 123-136, 2000.
13. Jürimäe, J, Mäestu, J, Jürimäe, T, and Pihl, E. Relationship between rowing performance and different metabolic parameters in male rowers. Med Sport (Roma) 52: 119-126, 1999.
14. Kramer, JF, Leger, A, Paterson, DH, and Morrow, A. Rowing performance and selected descriptive, field, and laboratory variables. Can J Appl Physiol 19: 174-184, 1994.
15. Mäestu, J, Jürimäe, J, and Jürimäe, T. Monitoring of performance and training in rowing. Sports Med 35: 597-617, 2005.
16. Mäestu, J, Jürimäe, J, and Jürimäe, T. Prediction of 2000m rowing ergometer performance from metabolic and anthropometric variables in male rowers. Acta Kinesiologiae Universitatis Tartuensis 4: 199-208, 1999.
17. Marfell-Jones, M, Olds, T, Stewart, A, and Carter, L. International standards for anthropometric assessment. ISAK: Potchefstroom, 2006.
18. Nielsen, T, Daigneault, T, and Smith, M. FISA Coaching Development Programme Course. Lausanne: FISA, 1993.
19. Nolte, V. Rowing faster. Champaign, IL: Human Kinetics, 2005.
20. Patton, JF, Kraemer, WJ, Knuttgen, HG, and Harman, EA. Factors in maximal power production and in exercise endurance relative to maximal power. Eur J Appl Physiol 60: 222-227, 1990.
21. Riechman, SE, Zoeller, RF, Balasekaran, G, Goss, FL, and Robertson, RJ. Prediction of 2000m indoor rowing performance using a 30s sprint and maximal oxygen uptake. J Sports Sci 20: 681-687, 2002.
22. Russell, AP, Le Rossignol, PF, and Sparrow, WA. Prediction of elite schoolboy 2000 m rowing ergometer performance from metabolic, anthropometric and strength variables. J Sport Sci 16: 749-754, 1998.
23. Schabort, EJ, Hawley, JA, Hopkins, WG, and Blum, H. High reliability of performance of well-trained rowers on a rowing ergometer. J Sport Sci 17: 627-632, 1999.
24. Secher, NH. Physiological and biomechanical aspects of rowing. Implications for training. Sports Med 15: 24-42, 1993.
25. Secher, NH. Rowing. In: Endurance in sport. Shephard RJ and Astrand PO, eds. Oxford: Blackwell Science, 2000. pp. 836-843.
26. Shephard, RJ. Science and medicine of rowing: a review. J Sport Sci 16: 603-620, 1998.
27. Soper, C and Hume, PA. Reliability of power output during rowing changes with ergometer type and race distance. Sports Biomech 3: 237-248, 2004.
28. Tanner, JM. Human growth and constitution. In: Human biology-An introduction to human evolution, variation, growth and ecology. Harrison, GA, Weiner, JM, Tanner, JM, and Barnicot, NA, eds. Oxford: Oxford University Press, 1983.
29. Wilcox, AR and Bulbalian, R. Running economy and race performance of male and female cross-country runners. Med Sci Sports Exerc 15: 108-113, 1983.
30. Womack, CJ, Davis, SE, Wood, CM, Sauer, K, Alvarez, J, Weltman, A, and Gaesser, GA. Effects of training on physiological correlates of rowing ergometry performance. J Strength Cond Res 10: 2334-2338, 1996.

elite rowers; exercise test; regression analysis; performance prediction

© 2009 National Strength and Conditioning Association