Share this article on:

Kinematic Determinants and Physiological Response of Cross-Country Skiing at Maximal Speed


Medicine & Science in Sports & Exercise: July 2009 - Volume 41 - Issue 7 - pp 1476-1487
doi: 10.1249/MSS.0b013e31819b0516
Applied Sciences

Purpose: The scope of the study was to a) transfer the maximal anaerobic running test (MART) to a treadmill-based cross-country roller skiing test protocol for skating and classic style, b) analyze the development and determinants of kinematic and physiological parameters from submaximal to maximal skiing speeds, c) analyze the effects of fatigue on skiing technique, and d) test the hypotheses that maximal skiing speed is related to cycle length and that the fastest skiers show shorter thrust combined with longer swing phases.

Methods: Up to 24 elite skiers performed MART tests in double poling, diagonal stride and V2 skating, with roller skis on a treadmill. Anthropometrics, blood lactate, HR, and kinematics were determined.

Results: When compared with former studies, faster skiing speeds of up to 34% were reported. Skiers improved speed by increasing cycle rate while trying to maintain cycle length. In the diagonal stride and V2 skating at maximal skiing speeds, skiers used contrary technical strategies to maintain the speed, whereas in double poling, a tendency toward an optimum in cycle length and cycle rate was established. Duration for the pole thrusts was around 180 ms for V2 skating and 210 ms for double poling. Highest relation to performance was found for the duration of the swing phases of arms and legs.

Conclusions: The increase in skiing speed and cycle length, compared with former studies and the positive relation between swing phase duration, stresses the importance of effective thrust phases within a short period. Therefore, it is recommended to increase the proportion of training aimed at the improvement of specific explosive strength and maximal power to increase the impulse of force.

1Department of Sport Science and Kinesiology, University of Salzburg, AUSTRIA; and 2Christian Doppler Laboratory "Biomechanics in Skiing," University of Salzburg, AUSTRIA

Address for correspondence: Thomas Stöggl, Ph.D., Department of Sport Science and Kinesiology, Christian Doppler Laboratory "Biomechanics in Skiing," University of Salzburg, Rifer Schlossallee 49 5400 Taxach/Rif, Austria; E-mail:

Submitted for publication August 2008.

Accepted for publication January 2009.

Biomechanical and physiological investigations of different cross-country skiing techniques across speeds and grades were the scope of several studies in the last decade. Most of these studies demonstrated that increases in submaximal speeds are associated either with increases in cycle rate with unchanged cycle length (11,15,25) or by increasing both (7). Regarding the change from submaximal to maximal skiing velocities, most studies (7,11,26) reported a decrease in cycle length, which is in contrast to maintained cycle length found by Nilsson et al. (15). Thus, there is a consensus regarding speed adaptation across a range of submaximal speeds with sole increase in cycle rate but a discrepancy concerning cycle length, especially when reaching maximal skiing speeds. In addition, to the best of our knowledge, with the exception of Bilodeau et al. (1), no study addressed the effects of fatigue on the aforementioned kinematic parameters. In particular, the sprint competitions are not only of special interest concerning how fatigue affects technique when approaching the finishing line but also which strategies might be applied to best compensate for it. It was noted that none of the studies analyzed the full spectrum across speeds, including the fatigued state.

Interestingly, in the studies above, neither of the analyzed variables cycle length or cycle rate was related to performance. In most studies analyzing relationships to performance, it was demonstrated that cycle length is a critical determinant of skiing velocity and/or race performance, especially for the diagonal stride and skating techniques (e.g., 1,2,20,24,26). Regarding the double poling technique, controversial results were reported showing on the one hand correlations of cycle rate but not of cycle length to performance (19) and on the other hand the opposite was reported (22,23). Bilodeau et al. (1) found only weak or no correlations for cycle length in double poling during race performance of a 50-km race. In addition, studies indicated that double poling performance can be improved by an increase in cycle length because of improved strength or muscle power factors (5,16) and/or technique developments (8,9). Although results are somewhat controversial, the general opinion is that a skier should focus on cycle rate as the mechanism of velocity control, whereas cycle length discriminates between skiers of different performance levels. Regarding the methods used in those studies, the procedures were both complex and time-consuming in the evaluation. Therefore, a quick feedback to athletes and coaches was hardly realizable with the applied methods.

Current studies report technical modifications applied principally during sprint races. In this same connection, recent analysis of a sprint double poling technique with higher peak pole forces and impulse of forces within a shorter poling time was conducted, which also required longer recovery phases (8) and the transfer of the "double-push" technique from inline skating to cross-country skiing (29). Both technical modifications were found to be superior to the conventional skiing style. In addition, the mean skiing velocity in World Cup sprint races in both the skating and classic techniques showed a steady increase in skiing speed (28), reaching mean race speeds of up to 9.5 m·s−1 in the classic and up to 10 m·s−1 in the skating styles. Here, it should be noted that skiing speeds in most of the above-mentioned studies are quite apart from these values. With the exception of Holmberg et al. (8,9), Bilodeau et al. (1), and Stöggl et al. (26,29), the analyzed maximal skiing speeds were much lower, falling in the range of 5.5 to 6.2 m·s−1 (7,11,15).

From a physiological perspective, primarily, the physiological response of different cross-country skiing techniques was analyzed (6,14). Oxygen uptake and work economy was found to be related to cross-country skiing performance (10,12). In a recent sprint study using the classic style (26), it was shown that faster athletes produced higher blood lactate levels with less decrease in their peak values over three heats. In contrast to those findings, it was shown that peak lactate and peak HR were not related to a single 1000-m double poling sprint bout (27). It also was demonstrated that short-duration maximal skiing speed and specific maximal strength are good predictors of cross-country skiing sprint performance. Continuous all-out tests conducted during a defined period or number of repetitions (16,27), or a short-duration ramp protocol up to maximal speeds (26), were used. Furthermore, the maximal anaerobic running test (MART) by Rusko et al. (21) was established as a successful diagnostic tool for determining both the metabolic and neuromuscular components of maximal anaerobic performance capacity of sprint and endurance athletes. This was true for sports such as running and cycling (which are comparable to cross-country skiing) (e.g., ]18,21[). Moreover, the MART concept offers the possibility of analyzing alterations of physiological and biomechanical variables across speeds in a single time-economic test.

Aims of the study were a) to transfer the MART to a treadmill-based cross-country roller skiing test protocol in both the skating and classic cross-country skiing techniques and b) to analyze the development and the determinants of kinematic and physiological parameters from submaximal to presently maximal possible skiing speeds, including the fatigued state of elite skiers in the skating and classic techniques. The main hypotheses in the present study were that, in the classic and skating techniques: 1) maximal skiing speed is related to cycle length, 2) the fastest skiers show shorter thrust phases combined with longer recovery phases in both the classic and skating styles, and 3) peak lactate is related to maximal skiing performance in all analyzed techniques.

Back to Top | Article Outline



Up to 24 (diagonal, n = 16; double poling, n = 22; V2 skating, n = 24) elite male cross-country skiers of the Austrian and Greek national teams (age 26 ± 4 yr, height 180 ± 6 cm, weight 75 ± 5 kg), including 7 skiers who were ranked among the top 20 in sprint or distance World Cup races and an Olympic champion, volunteered as subjects in the study, which was approved by the ethics committee of the University of Salzburg. All the skiers were fully acquainted with the nature of the study before they gave their written informed consent to participate. Each subject was accustomed to roller skiing on the treadmill at high speeds, in both the classic and skating style, from numerous training and testing sessions within a period of 2 yr. Because of injury and overloading syndromes within the testing phase or special affiliations for the skating and classic style, not all of the 24 subjects competed in all three techniques.

Back to Top | Article Outline

Skiing Techniques and Cycle Definitions

In this study, the following three cross-country skiing techniques were analyzed.

Back to Top | Article Outline

Double poling.

With this technique, all the propulsive forces were symmetrically and synchronously applied during the ground contact of the poles. One double poling cycle was subdivided into a poling phase starting at the pole plant of both poles and ending when the poles were no longer in contact with the ground, and a swing phase where the arms are repositioned for the next pole plant.

Back to Top | Article Outline

Diagonal stride.

In this technique, the arms and legs move in a diagonal fashion in which the arm push-off is performed together with the push-off of the leg on the contralateral side of the body. The arm movement was divided into a poling phase and a swing phase. In addition, the time between pole-out of one pole to the pole-plant of the contra arm was defined as pole noncontact phase. The movement of the legs was split up into a positioning phase starting with the leg switch and ending at the consecutive leg switch during ground contact. At high grades or speeds, skiers began to partially run or jump-no longer demonstrating a gliding phase. Hence, the preparation phase was calculated from set down of the roller ski to the next leg switch, immediately before leg thrust. The thrust phase was defined from leg switch to the time when the roller ski leaves ground contact. The swing phase was defined from the point where the roller ski leaves ground contact up to the next leg switch. In addition, the time difference between the end of leg thrust and the end of pole thrust of the contra side was calculated. The single-cycle variables are given in absolute and relative (% total cycle) values.

Back to Top | Article Outline

V2 skating technique.

The V2 technique has a symmetrical and synchronous pole push at each leg push-off. In addition, two modes of the V2 skating technique were reported: the conventional V2 and the "double-push" V2 skating technique. In both modes, the basic movement and leg/arm coordination pattern are equal. The difference lies in an additional push-off where the leg is adducted with respect to the body instead of the normally static glide period in the conventional V2 (29). For the V2 technique, one cycle was defined from one pole plant to the next pole plant. This represents a half cycle when regarding only the leg work. The movement of the arms is equal to the double poling technique. The thrust phase of the leg movement starts with pole plant and ends when the roller ski leaves ground contact (29). Leg-closing phase was defined from end of leg-thrust up to the point where the leg closes toward the other leg. The precise instant of the set down of the foot was not analyzed because of an insufficient side view on the whole roller ski. To analyze arm and leg coordination patterns, two calculations were performed: first, the time difference between the end of the leg-thrust phase to the end of poling phase, and second, the time difference between leg closing and pole plant of the arms.

Back to Top | Article Outline

Overall Design of the Study

Testing was performed within 2 d, with at least 48 h in the period of spring and beginning of summer. The first day started with a 15-min warm-up using the classic style, followed by a 5-min break, and finally, the MART test in double poling. A rotation system was used starting with the warm-up of the next subject when the test in double poling of the first subject was done. Consequently, each subject had a break of 35 min to the pending MART in the diagonal stride. In the break between tests, each subject performed a 10-min cool-down while running on a treadmill and a 5-min warm-up protocol on roller skis immediately before start of the second MART. All subjects had the same test order on day 1, starting with the MART in double poling and followed by the MART in the diagonal stride. The second day consisted of a 15-min warm-up in the skating technique and the MART test using V2 skating. Activity and break times in between the single tests were controlled and standardized. During the breaks, the subjects were allowed to drink a standard fluid (mixture of carbohydrates, proteins, and electrolytes) that was provided by the testing team.

Back to Top | Article Outline

Testing Protocols


The two warm-up sessions on the treadmill in the classic and skating styles, respectively, lasted 15 min and included double poling, diagonal stride, and kick-double poling for classic and V1, V2, and V2 alternate sections with increasing intensity. Two parts in the classic warm-up contained a steady increase of speed up to submaximal speeds in double poling and diagonal stride, respectively. During the skating warm-up, three parts of increasing speed up to submaximal level were performed for V1, V2, and V2 alternate, respectively. Those speed increases were included to prepare the athletes for the forthcoming physiological and coordinative demands at higher roller skiing velocities in the single tests.

Back to Top | Article Outline

MART protocols.

On the basis of the test protocol of the MART of Rusko et al. (21), three protocols for double poling, diagonal stride, and V2 skating using roller skis on a treadmill were designed. The protocol included stages of 30 s with 100-s break in between. In double poling and V2, treadmill speed was increased at 0.3-m·s−1 increments starting with 7 m·s−1 (inclination double poling = 1.5°, V2 = 2.5°). For the diagonal stride, treadmill inclination was increased 1° every stage, starting at 6° with constant treadmill speed of 4.5 m·s−1 (the reason for a grade protocol in the diagonal stride is based on test experiences of the last 4 yr, showing that, in speed protocols, roller skis were noticeably difficult to handle especially at high treadmill speeds). The test was stopped by the tester when the subject passed a marker placed 1.5 m ahead from the front of the treadmill. Maximal performance (Vmax) in double poling and V2 skating was calculated by linear interpolation using the following formula: Vmax = Vf + ((t / 30) 0.3 m·s−1), where Vf is the velocity of the last completed workload (m·s−1), t is the duration of the last workload (s), and 0.3 m·s−1 is the velocity difference (ΔV) between the last two workloads. For the diagonal stride, the formula was as follows: Grademax = Gradef + ((t / 30) × 1°), with Gradef as the grade of the last completed stage. Lactate and blood glucose were taken in the 60 s of each 100-s rest period, and in the first, third, fifth, and seventh minutes after the end of the last stage. For determination of peak HR of every stage, the highest value at the end or at the first second after the end of the stage was taken. In addition, for the determination of HR recovery within each break, the HR value just before start of the next stage was recorded. All stages were video analyzed while five consecutive cycles starting around the 10th second of each stage were taken. One cycle in each technique was defined from the time between two subsequent pole plants of the right arm, whereas additional subdefinitions were performed especially for the diagonal stride and V2 skating technique. To determine the effects of fatigue on kinematic parameters, the aforementioned five cycles at the beginning of the final stage were compared with the last five cycles just before test termination.

Back to Top | Article Outline


Roller skis for the tests using the classic style were Pro-Ski C2 (Sterners, Nyhammar, Sweden), and Swenor Skates (Norway) were used for the skating tests. Each subject used the same pair of roller skis, which were warmed up before beginning the testing session by roller skiing for 20 min on the treadmill. The purpose of this session was to prevent a warm-up effect of the wheels and bearings during testing. All roller ski tests were performed on a large treadmill (Pomer; Wiege Data, Leipzig, Germany; belt dimensions of 4.5 × 3 m), on which roller skis could easily be used. The belt of the treadmill consisted of a nonslip rubber surface, allowing subjects to use their own poles with special carbide tips. Pole length for the classic style was 147 ± 5 cm (82 ± 1% of body height) and 158 ± 6 cm (88 ± 1%) for the skating technique. Before the start of each treadmill test, the athlete was secured with a safety harness, which was connected to an emergency brake suspended from a metal bracket above the treadmill. Mean air temperature was 18 ± 1°C, and the mean relative humidity was 54 ± 7% for all days of testing. HR was measured throughout the tests by an HR monitor (Suunto T6, Helsinki, Finland) using a 2-s interval for data storage. Blood samples (20 μL) were taken from the hyperemized earlobe for the determination of blood lactate and blood glucose (Biosen S_line; EKF-Diagnostics GmbH, Magdeburg, Germany). The analyzer was calibrated before each test and checked using a lactate standard of 12 mmol·L−1. Calibration results within ±0.1 mmol·L−1 were accepted. For two-dimensional video analysis, a video camera (50 Hz; Sony, Tokyo, Japan) was fixed at the rear end of the treadmill with a total view on the frontal plane of the subject and the whole movement range of the poles, to determine kinematic cycle characteristics during each test in each technique. Cycle length was determined by multiplication of cycle time times treadmill velocity (cycle length = cycle time × velocity). The kinematic analysis of cycle characteristics was done using Dartfish Pro Suite software (Dartfish Video Software Solutions, Fribourg, Switzerland).

Back to Top | Article Outline

Statistical Analyses

All data were checked for normality, calculated with conventional procedures, and presented as means ± SD. The course of selected variables across all velocity/grade stages will be illustrated in Figures 1, and 2 and in Tables 1-4. For comparisons across speeds and for correlation analysis toward maximal performance, the stages that were successfully performed by all subjects (first four stages for all three techniques) together with the individual maximal stage were taken. Changes across speeds were analyzed using repeated-measures ANOVA applying Bonferroni α correction. For determining the relationships between measured variables and performance in a first step, a trend analysis was performed to achieve the best-fit regression model restricted to linear and polynomial (second and third order) models. In the case of linear relationships, the Pearson product moment correlation coefficient was calculated. For nonlinear relationships, the best-fit regression model was taken, and the coefficient of determination (R2) was presented. To determine the effects of fatigue on kinematic variables, the first four cycles taken 5 s after the end of the acceleration phase were compared with the last four cycles just before test conclusion. Paired-sample t-tests were applied for this analysis. The statistical level of significance was set at P< 0.05 for all analyses. All statistical tests were processed using SPSS 15.0 Software (SPSS, Inc., Chicago, IL) and Office Excel 2003 (Microsoft Corporation, Redmond, WA).

Back to Top | Article Outline


Back to Top | Article Outline

Descriptive analysis.

Maximal roller skiing speeds and grade, respectively, ranged from 7.75 to 8.74 m·s−1 (8.17 ± 0.3) for double poling, from 7.83 to 9.30 m·s−1 (8.90 ± 0.30) for V2, and from 8.93° to 11.87° (11.03 ± 0.96) for the diagonal stride. Peak lactate, peak blood glucose and peak HR values were 10.5 ± 2.5 mmol·L−1, 121 ± 19 mg·dL−1, and 183 ± 6 bpm for double poling; 14.5 ± 3.4 mmol·L−1, 115 ± 16, and 184 ± 7 bpm for the diagonal stride; and 12.8 ± 2.8 mmol·L−1, 125 ± 18, and 184 ± 8 bpm for V2 skating, respectively. Peak lactate was significantly higher (P < 0.001) in the diagonal test compared with the other two techniques, whereas the lowest values were found in double poling. HR and blood glucose levels were equal in all three techniques. Roller skiing speed at a lactate value of 4 and 10 mmol·L−1 were 6.74 ± 0.46 and 8.22 ± 0.25 m·s−1, respectively, for DP, and 7.26 ± 0.44 and 8.75 ± 0.37 m·s−1, respectively, for V2 skating. For the diagonal stride, the inclination at the lactate values of 4 and 10 mmol·L−1 was 5.84 ± 1.15° and 10.10 ± 0.69°, respectively. The development of all measured physiological data across speeds or grades, respectively, is presented in Table 1.

Back to Top | Article Outline

Development of parameters across speeds.

Detailed information about all measured kinematic parameters across speeds in the three techniques is presented in Tables 2 and 3. In all three techniques, athletes adapted to increasing intensity by a decrease in poling and swing time of the arms and an increase in cycle rate (P < 0.01). At the highest skiing speed, poling time was 210 ± 12 ms in double poling (at 8.8 m·s−1) and 176 ± 12 ms in the V2 skating style (at9.4 m·s−1; Fig. 1). The calculated covered distance during the thrust phases of the poles (d = poling time × speed) remained constant across speeds for the double poling technique (∼1.83 ± 0.12 m) and started to decrease in the V2 technique from the highest submaximal level that all skiers were able to perform to individual maximal speeds (from 1.74 to 1.63 m, P < 0.001). Cycle length remained constant for the first two stages and, afterward, started to decrease with increasing intensity up to the maximal intensity in double poling and the diagonal stride (P < 0.01). In V2 skating, cycle length indicated an insignificant increase up to the fourth stage, and thereafter decreased up to maximal speed (P < 0.01). However, cycle length at maximal speed was not significantly different to the values of the first two stages (Fig. 2). Regarding the relativized values (% cycle) of poling time, the values remained constant for the first stages and increased toward maximal speed in double poling and the diagonal stride (both P < 0.01). The proportion of poling time during the entire cycle was unchanged in the V2 technique across speeds. For relative swing time, the inverse pattern is considered. In the diagonal stride, a phase where none of the poles had ground contact was observed. This time gap decreased from 0.19 ± 0.04 s down to 0.08 ± 0.04 s at maximal intensity (P < 0.001).

The thrust phase of the legs in V2 skating remained constant for the first two stages and decreased toward maximal speed (P < 0.001). In the diagonal stride, the duration of the thrust phase remained constant across all grades; however, swing time of the legs and preparation phase began to decrease after the second stage (both P < 0.01). The duration for leg closing in V2 skating remained constant at submaximal speeds and actually decreased toward individual maximal speeds (P < 0.001). In both the diagonal stride and the V2 skating style, the push-off of the legs was terminated later in time than the pole thrust and showed no change across speeds. The instant of leg closing in V2 skating and the instant of pole plant occurred, on average, 0.07 s earlier (mean value across speeds) indicating no significant time difference across speeds. Thirteen skiers switched to the double-push skating technique at a skiing speed of approximately 8.2 m·s−1. The group of double-push skiers demonstrated higher maximal speeds compared with those using the conventional style (9.02 ± 0.12 vs 8.77 ± 0.4 m·s−1; P<0.05). Furthermore, these particular skiers showed higher cycle lengths across all analyzed speeds (mean over all speeds was 7.3 ± 0.3 vs 6.47 ± 0.2 m, P < 0.01).

Back to Top | Article Outline

Effects of fatigue.

The effects of fatigue on kinematic parameters are illustrated in Tables 2 to 4. In a fatigued state, cycle rate increased in double poling (P < 0.001) from 1.17 ± 0.16 to 1.31 ± 0.16 Hz, in the V2 from 1.40 ± 0.10 to 1.48 ± 0.09 (P < 0.01), and was unchanged in the diagonal stride. In contrast, cycle length decreased in double poling from 7.23 ± 0.99 to 6.79 ± 0.80 m (P < 0.05), in the V2 from 6.47 ± 0.57 to 6.09 ± 0.51 m (P < 0.001), and was unchanged in the diagonal stride. Poling time remained constant for all three techniques, whereas the swing phase of the arms decreased in double poling from 0.65 ± 0.12 to 0.55 ± 0.08 s (P < 0.001), in the V2 from 0.54 ± 0.05 to 0.49 ± 0.04 s (P < 0.01), and in the diagonal stride from 0.48 ± 0.10 to 0.42 ± 0.07 s (P < 0.05). In percentage of the total cycle, the relative poling time increased in DP from 25.4 ± 3.4% to 29.1 ± 2.4%, for V2 from 25.3 ± 2.1% to 27.1 ± 2.2% (both P < 0.001), and remained constant for the diagonal stride. For the movement pattern of the legs, it was shown that, in a state of fatigue, the duration of the leg thrust increased for the diagonal stride (0.15 ± 0.04 to 0.22 ± 0.09, P < 0.05) and remained constant for the V2 skating technique, but increased from 21.2 ± 1.5% to 22.9 ± 3.2% (P < 0.05) when expressed in percent of cycle time. In addition, it was noted that the time for leg closing in V2 decreased from 0.35 ± 0.04 to 0.30 ± 0.08 s (P < 0.05). For the diagonal stride, the leg swing time increased from 0.28 ± 0.04 to 0.41 ± 0.08 s; however, in contrast, the preparation time decreased from 0.39 ± 0.12 to 0.08 ± 0.18 s (both P < 0.001).

Back to Top | Article Outline

Correlations to maximal performance.

Only for double poling the anthropometrical variables body height (r=0.46, P < 0.05) and pole length (r=0.42, P < 0.05), indicated low correlations to maximal speed; however, when relativizing pole length to body height, no further correlation remained. Regarding physiological data, only peak lactate showed correlations to the respective maximal speed in all techniques (r = 0.83 for diagonal, r = 0.67 for double poling, both P < 0.001; and r = 0.49 for V2, P < 0.05). Performance at the 10-mmol·L−1 lactate value was related to double poling and V2 skating performance (both r = 0.43, P < 0.05). At submaximal stages, cycle rate demonstrated negative correlations, whereas cycle length showed positive correlations to maximal performance in all three techniques. At the individual maximal intensity stage, only in the V2 skating technique did cycle length positively correlate toward maximal speed (Tables 5 and 6). However, when regarding absolute skiing speed/grade near the maximal reached values that in the diagonal stride (12°, N = 9) and the V2 skating technique (9.1 m·s−1, N = 16), no correlation of cycle length to maximal skiing speed/grade was found (Fig. 3). With the speed increasing, a nonlinear quadratic relationship toward maximal performance was observed in the double poling technique (8.2 m·s−1, R2 = 0.62; 8.5 m·s−1, R2 = 0.73; individual maximal speed, R2 = 0.48), which indicated optimal cycle rate and cycle length of around 1.2 Hz and 7.5 m, respectively (Figs. 3 and 4). When observing arm movement, only the swing time showed positive correlations toward maximal performance. A comparison of cycle time values suggested that relative poling time was negatively correlated to maximal performance, whereas relative swing time positively correlated to the same. For the diagonal stride, both the duration of the swing time of the leg and the duration of preparation phase (both at submaximal levels) were positively related to maximal performance. For the V2 skating technique, the duration of the thrust phase showed negative correlations toward maximal speed at submaximal and maximal speeds. In addition, the duration for the leg closing phase at submaximal speeds showed positive relations. The time difference between the point of leg closing and pole plant at submaximal levels was inversely related to performance. Because of the negative values of this parameter, a longer time between leg closing and the pending pole plant seems to be advantageous.

Back to Top | Article Outline


This study provides additional information on kinematic and physiological data of present-day elite skiers who performed at submaximal and maximal skiing speeds in both the classic and skating technique. The key findings of the study are as follows:

1) The transfer of the MART test from running and cycling (21) to a treadmill-based roller ski protocol was possible for all three cross-country skiing techniques. Moreover, the testing concept, together with the usage of a simple two-dimensional video method, allowed a kinematic analysis of variables across speeds up to maximal performance.

2) Skiers increase speed in all techniques by increasing cycle rate while trying to maintain cycle length.

3) At maximal skiing speeds in the diagonal stride and V2 skating, skiers are using somewhat contrary technical strategies to keep the speed, whereas in the double poling technique, a tendency toward an optimum in cycle length and cycle rate was established.

4) The highest relation to performance was found for the duration of the recovery/swing phases stressing effective thrust phases at limited time durations.

The constant rise in cycle rate in double poling, diagonal stride, and V2 skating in the present study was in line with previous research (7,11,15,26) and was proposed as the primary method of increasing speed (22). What is more, the maintained cycle length at submaximal speeds using double poling agrees with the findings of earlier studies (7,11,15). Interestingly, when reaching maximal speeds, Nilsson et al. (15) reported no change, whereas Millet et al. (11) and Hoffmann et al. (7) found a decrease in cycle length in double poling that is in line with the results attained by the present study. Further evidence of this decrease in cycle length was noted in the study by Stöggl et al. (26), which indicated that cycle length showed a steady decrease up to maximal speed. On the one hand, this effect might be related to a different test protocol where a continuous ramp test was applied that is in contrast to the incremental protocol in the present study and the studies mentioned above. On the other hand, it might be the result of the smaller range between submaximal and maximal speeds.

Regarding the V2 technique, an insignificant rise in cycle length at the first stages followed by a significant decrease toward maximal speeds was established. However, no difference was found between the highest submaximal and the maximal speeds. This result is again comparable with the findings of Millet et al. (11) where a significant increase in cycle length at submaximal speeds followed by a decrease toward maximal speed was found. Nilsson et al. (15) also reported unchanged cycle lengths across all measured speeds. Obviously, there is a consensus of most studies regarding adaptation in cycle rate and cycle length during double poling and V2 skating, although the investigated skiing speeds in the current study were much higher.

For the diagonal stride, a steady decrease starting at the second intensity stage toward maximal performance was found. In contrast, Nilsson et al. and Stöggl et al. (26) did not report a change in cycle length across speeds. Here, it is worth noting that, in the present study, a grade protocol was used, which is in contrast to speed protocols on flat terrain (15) and at an 8° inclination (26). Hence, it might be speculated that, in the diagonal stride, skiers adapt to an increase in speed at constant grade exclusively by an increase of cycle rate while maintaining cycle length. However, when grade is altered at constant speed, skiers no longer seem to be able to hold their cycle length. This might be attributed to a changed mechanical situation when grade is increasing; for example, that the gravitational force component that is working against forward direction along the grade (FH) increases with grade according to the following equation: FH = FG sinα (with FG = gravitational force, α = grade). This gravitational force component acts on the body throughout the cycle and increases with grade. This is especially true during the cycle phase when no propulsion will be applied, namely, the swing phase for the legs together with the preparation phase and the pole noncontact phase. This also can be seen in the significant decrease in swing time of both the legs and the arms and in the reduction of time for preparation, whereas push-off time of the legs remains constant and poling time shows only a small decrease in maximal speeds. Consequently, to compensate for this sine-related increase in FH, skiers would have to increase leg and pole propulsion in every step at a level that prohibits the reduction of the passive swing phases-which seems not to be possible in that group of skiers.

Regarding cycle characteristics, it should be mentioned that the skiers in the current study were analyzed at skiing speeds that were approximately 30% to 45% higher in double poling, 24% to 44% higher in V2 skating compared with recent studies (7,11,15), and around 3% lower than was seen in the study by Stöggl et al. (26) for double poling. In the diagonal stride, the comparison is somewhat difficult because of the different protocols. However, the basic speed of 4.5 m·s−1 at ∼11.3° was 7% slower compared with the findings of Stöggl et al. (26) at a lower grade (8°), and 27% slower when compared with Nilsson et al. (15) on flat terrain. In that context, it is interesting how skiers are able to show a comparable cycle rate and cycle length pattern to earlier studies at skiing speeds of up to 45% higher. Regarding cycle length at maximal skiing speed in double poling, the skiers were able to produce between 31% (11,15,26) and 75% (7) longer cycle length (around 1.7-3 m longer) in the current study at approximately the same or lower cycle rate. The same could be found for the V2 skating technique where the cycle length was approximately 18%-30% longer (around 2-3 m longer) and cycle rate approximately 0.1 Hz higher compared with earlier studies (11,15). These results, when compared with those of earlier studies, indicate that the increased cycle length provides the basis for the achievement of high skiing speeds, particularly in the double poling and V2 skating technique.

At this point, the question arises, "How are skiers able to produce up to 75% longer cycle lengths at up to 45% higher skiing speeds in double poling and the V2 skating technique?" In addition to this question, it should be considered that, at maximal skiing speeds, the time for propulsion via the poles showed values of only 0.21 s in double poling and 0.18 s in V2 skating. These time values are in the neighborhood of very short contact times in jumping exercises (e.g.,ground contact time for a drop jump) and being around 100 ms shorter as reported in earlier studies that analyzed both techniques at maximal skiing speeds (8,11,15). An answer to the question above might be found in modified skiing techniques in use during the last years. Regarding the V2 skating technique, Stöggl et al. (29) investigated the effectiveness of the double-push technique on skis, transferred from inline speed skating, on maximal skiing speed on a slightly uphill terrain. The double-push technique proved to be 2.9% faster on a 100-m track compared with the conventional V2 technique. In addition, it was found that the skiers were able to ski at longer cycle lengths and lower cycle rates when using this new technique. In the present study, 13 skiers used the double-push technique when speeds approached maximal levels, which resulted in higher maximal speeds when compared with the speeds of skiers using the conventional technique. Holmberg et al. (8) reported on a sprinter-like double poling strategy, where skiers produced higher peak pole forces, higher impulse of forces, and showed higher maximal speeds. This more explosive double poling technique led to shorter relative poling times and longer relative recovery phases. As a result, they made a shift to a shorter time of high activation and a longer time for recovery. In addition, the longer recovery phase and lower cycle rate of this modern form of double poling was shown to be related to a lower HR response and blood lactate concentration and to longer time to exhaustion (9). Furthermore, as to the shorter poling times in the current study, absolute (11,15) and relative (8,11,15) recovery time were even longer when compared with earlier studies at lower speeds. Hence, it might be speculated that the technical development in double poling was already adopted and further developed by the currently analyzed elite skiers, as was suggested by Holmberg et al. (8,16). From a training perspective, the decrease of poling time to 210 ms in double poling and 180 ms in V2 stresses the importance of high force development and force production within a very short period. All these aspects should be considered in modern strength and technique training concepts of cross-country skiers. Therefore, some studies already support this argument showing an improved double poling performance without any increase in cycle rate, a result of improved strength or muscle power and/or technique development (5).

Regarding the development of cycle rate, cycle length, and thrust times in all techniques up to maximal speed/grade, one might ask, "Where can the limitations and possibilities for further increases in performance in the single techniques be found?" One major problem is the constraint of stopping the foot or the pole with respect to the ground, as is the case for the pole thrusts in all techniques and the leg thrust in the diagonal stride. Therefore, the possible duration to apply force during the thrust phase is inversely related to skiing speed. For example, in the double poling technique, a poling time of 210 ms was found at a treadmill speed of 8.8 m·s−1; consequently, the pole had ground contact over a calculated distance of 1.85 m. If we presume that the covered distance remains constant across speeds, as was found in the current study, and a skier should ski at a velocity of 10 m·s−1, the maximal possible poling time would be 185 ms. By doing so, the ability to store and release mechanical energy to and from the elastic structures during a decreasing duration provides a limiting factor. Also consider that muscle contraction generates less force for propulsion because it is directly related to progression speed (13). In this context, it could be argued that these short absolute times represent a critical limit for the full recruitment of both Types I and II fibers for sufficient force generation. Studies on force-velocity characteristics regarding contraction velocity on single fibers in vivo demonstrated that Type I fibers need 100-140 ms and Type IIA fibers need ∼55-85 ms to create maximal tension and thereby force (e.g., ]4[).

Possible ways to further increase maximal skiing speed might be found in four aspects: 1) to increase the covered distance of the thrust phase of the arms, which might be fulfilled by longer poles, but is limited regarding the position of set down and release of poles from the ground. Because pole angle at pole plant should be at least 90° with respect to the ground in order not to contribute to a braking effect, the distance might be prolonged by a larger forward lean of the entire body paralleled by a higher position of the center of mass in the form of a forward jump; 2) to further increase cycle rate: although here, too, a critical limit might be given from a mechanical and physiological point of view; 3) to increase the impulse of force during those short thrust phases by an increased explosive and maximal strength level; and 4) to consider Minetti's (13) recommendation of searching for a new pole design that would allow the pushing portion of the pole to slide. This way, muscle contraction speed would be much lower while continuing to slide on the medium. In that context, it might be added that in contrast to the diagonal stride and double poling, no decrease in cycle length across speeds was found for the V2 skating technique, where a sliding push-off in contrast to a fixed one with respect to the ground is performed.

Regarding relationships between kinematic variables and maximal performance, a consensus to former literature could have been found when regarding cycle length as a predictor of performance. However, for double poling and diagonal stride, this positive relationship was found only at submaximal levels, whereas when reaching maximal speeds/grades, neither cycle length nor cycle rate was correlated to maximal speed. Regarding double poling, one might consider that, with increasing speeds, a curvilinear relationship showed better fitting to the cycle length-maximal speed plot than that of a linear relationship (Figs. 3 and 4). For double poling, there seems to be an optimum in the cycle length and cycle rate pattern to achieve maximal performance, keeping in mind that at maximal speed, values of around 1.2 Hz in cycle rate and 7.5 m of cycle length seemed to be optimal to achieve maximal high skiing speeds. Regarding the diagonal stride and the V2 skating technique, the correlation of cycle length at submaximal speeds vanished when reaching absolute speeds near the maximal achieved skiing speeds (Fig. 3). However, at the individual maximal speed in V2 skating, faster skiers were able to produce longer cycle lengths, although they had to manage higher treadmill speeds. It was concluded that at high skiing speeds/grades skiers applied individual strategies that were, to a degree, opposite to those of others. For instance, in the diagonal stride, some skiers tried to concentrate on long cycle length, whereas others started to show more of a running diagonal stride with high cycle rate and no gliding phase. The same observation was found in the V2 skating technique, where 13 skiers started to use the double-push technique (29) when speed reached high values, whereas others tried to increase cycle rate to hold the speed. This characteristic might be compared with other sports where new technical developments emerged. For instance, when observing athletes performance in ski jumping when the V technique emerged and in alpine skiing with the development of carving skis and shorter ski lengths, there evolved a gap in performance between skiers using the old style and those using the new style. Some athletes adapted quickly to the new techniques and modified material, whereas others had problems initially. However, over the years, all skiers became more comfortable with the new conditions and/or equipment that were found to be superior to the old styles and the gap in performance was closed again. In double poling, it might be speculated that the optimum style has already been discovered; however, in the two other techniques, the future will show which strategies might be the most successful.

When regarding the subdivisions of the cycle, it can be found that the duration of the swing phase of the arms was positively related to performance in all techniques. Concerning the relative values for poling and swing, the proportion for the swing phase over the total cycle showed a positive correlation, whereas in the duration of the thrust, a negative correlation. In addition, the duration of the swing phase of the legs in the diagonal stride at submaximal speeds, and the duration for leg closing at both submaximal and maximal speeds in V2, positively correlated to performance. Only in V2 skating was the absolute duration of the thrust phase of the legs inversely related, pointing to a faster push-off of better-performing skiers. In summary, these results once more stress the effect of swing phases of longer duration, which, in a way, might be regarded as recovery phases-where muscles are able to quickly switch off (more or less). Of note is that a longer recovery phase has been shown to be one criterion for a modern and fast DP strategy in elite cross-country skiers with lower blood lactate and HR (8,9). Combined with the results concerning the longer cycle lengths and recovery times, this outcome again stresses the high importance of an effective thrust phase to achieve those longer swing times of both the legs and the arms. The package of short thrust phases with a high impulse and, consequently, the possibility for swing or recovery phases of longer duration seems to attribute expressly at submaximal levels to a more economic and faster technique.

The negative correlation of the time difference between leg closing and the pending pole plant suggests that faster skiers displayed a larger time gap between the point of leg closing and the next pole plant. Accordingly, faster skiers showed both a longer duration of leg closing and a later start of pole plant representing the time point of the beginning of the leg thrust. Subsequently, those skiers demonstrated a longer gliding time where the center of mass is in a central position over the gliding skis.

About the measured physiological parameters, only peak lactate showed positive relations to maximal performance. The speed/grade at 10 lactate level simply showed weak correlations. The results are somewhat weaker compared with studies of the MART in running, where the power at 10mmol·L−1 lactate and peak lactate showed high correlations to 400-m time and maximal performance in the MART. Other studies (3,17) also found correlations between peak lactate and 400-m running times, suggesting that peak lactate can be used as a rough estimate of anaerobic capacity. These results suggest that high anaerobic capacity, in the context estimated by peak lactate (30), seems to be closely connected to cross-country skiing sprint performance of sprint skiers. HR, HR recovery between each stage, and blood glucose levels were not related to performance; which might cause one to question the use of these parameters for diagnostics to predict performance. Regarding anthropometric data only for double poling, a small tendency favoring taller skiers was observed. Parallel with the positive relationship of body height, the absolute pole length also was positively related to performance. However, when relativized on body height, the relationship vanished.

In both the double poling and the V2 skating techniques, skiers had to increase their cycle rate; however, cycle length decreased when they came into a state of fatigue moments before test termination. The time for the pole thrust in both techniques and leg thrust in V2 remained constant, whereas the swing time decreased. This pattern led to an increase in relative thrust time and a decrease in relative swing time. This result might be due to fatigue causing an ongoing decrease in effectiveness of the thrust phase, which leads to a reduction of the nonpropulsive swing phases. Therefore, the reduction in the forward thrust could only be compensated by an increase in cycle rate at the expense of a shortened recovery phase. Owing to assumed biomechanical and physiological limitations in a further increase of cycle rate, skiers were consequently no longer able to hold the treadmill speed. However, these findings should be further investigated by complex biomechanical and physiological measurements, which provide a detailed look at (and thereby aiding in the analysis of) the development of pole and leg forces together with physiological data during ongoing fatigue. Regarding the diagonal stride, no significant change in cycle rate, or cycle length was observed in a fatigued state. Although the time for the leg thrust and swing time increased, the time for preparation was close to zero. This result is because skiers started to jump or run instead of gliding. Accordingly, the time required for the preparation phase to be initiated from the set down of the roller ski to the next leg switch was sharply reduced. This reduction was a result of the elimination of the gliding phase, which had been defined as an element of the preparation phase. In the conventional stride, the preparation phase was defined with the first leg switch that occurs at the same time as the set down of the roller ski in this group of skiers and ends with the consecutive leg switch.

Back to Top | Article Outline


In summary, the present study shows that elite skiers controlled speed by an increase in cycle rate while at the same time aiming on maintaining cycle length in all three techniques. Cycle length was related to maximal performance only at submaximal levels, whereas at individual maximal speeds, neither cycle rate nor length showed linear relations to maximal speed for the classic techniques. It was evident that skiers used individual, somewhat contrary patterns to manage the highest achieved absolute maximal speeds. That is to say, skiers using the skating technique applied the double-push approach with longer cycle lengths, whereas others primarily aimed for a high cycle rate. For the diagonal stride, a high-frequency running/jumping technique with no gliding phase was countered by the goal to maintain long cycle lengths. In double poling, a pattern to an optimal cycle rate of approximately 1.2 Hz at a cycle length of 7.5 m was found. It is notable that the functional significance of the swing/recovery phases for both legs and arms was particularly important at submaximal speeds, whereas the thrust durations were partly negatively related. The strategy toward high activation within short thrust phases and consequently longer phases of swing and recovery especially at submaximal speeds seem to be highly related to performance. This strategy needs a well-developed ability to produce great forces for a limited time for the thrust, emphasizing the demands for specific explosive strength and highly developed motor skills. In a fatigued state, skiers were no longer able to maintain their cycle length at cost of a reduced swing time and consequently a forced increase of cycle rate to keep the speed. In summary, the applied method and test concept allowed an affordable (excluding the cost of a treadmill), easily applicable diagnostic tool for kinematic and physiological analysis, as well as for technique control at maximal skiing speeds. The illustration of technical tendencies and possible future developments, together with the selection of useful parameters, might provide useful aspects for coaches, athletes, and scientists.

No funding was received for this work from NIH, Wellcome Trust, HHMI, and others.

The authors would like to express appreciation for the support of Donna Kennedy, University of Salzburg, Austria. The authors also thank the athletes and trainers for their participation, enthusiasm and cooperation in this study. The results of the present study do not constitute endorsement by ACSM.

Back to Top | Article Outline


1. Bilodeau B, Rundell KW, Roy B, Boulay MR. Kinematics of cross-country ski racing. Med Sci Sports Exerc. 1996;28(1):128-38.
2. Boulay RM, Rundell KW, King DL. Effect of slope variation and skating technique on velocity in cross-country skiing. Med Sci Sports Exerc. 1994;27(2):281-7.
3. Fujitsuka N, Yamamoto T, Ohkuwa T, Santo M, Miyamura M. Peak blood lactate after short periods of maximal treadmill running. Eur J Appl Physiol. 1982;48(3):289-96.
4. Garnett R, O'Donovan MJ, Stephens JA, Taylor A. Evidence for the existence of three motor unit types in normal human gastrocnemius [Proceedings]. J Physiol. 1978;280:65P.
5. Hoff J, Helgerud J, Wisloff U. Maximal strength training improves work economy in trained female cross-country skiers. Med Sci Sports Exerc. 1999;31(6):870-7.
6. Hoffman MD, Clifford PS. Physiological responses to different cross country skiing techniques on level terrain. Med Sci Sports Exerc. 1990;22(6):841-8.
7. Hoffman MD, Clifford PS, Bender F. Effect of velocity on cycle rate and length for three roller skiing techniques. J Appl Biomech. 1995;11(3):257-66.
8. Holmberg H-C, Lindinger S, Stöggl T, Eitzlmair E, Müller E. Biomechanical analysis of double poling in elite cross-country skiers. Med Sci Sports Exerc. 2005;37(5):807-18.
9. Holmberg HC, Lindinger S, Stöggl T, Bjorklund G, Müller E. Contribution of the legs to double-poling performance in elite cross-country skiers. Med Sci Sports Exerc. 2006;38(10):1853-60.
10. Mahood NV, Kenefick RW, Kertzer R, Quinn TJ. Physiological determinants of cross-country ski racing performance. Med Sci Sports Exerc. 2001;33(8):1379-84.
11. Millet GY, Hoffman MD, Candau RB, Clifford PS. Poling forces during roller skiing: effects of technique and speed. Med Sci Sports Exerc. 1998;30(11):1645-53.
12. Millet GY, Perrey S, Candau R, Rouillon JD. Relationships between aerobic energy cost, performance and kinematic parameters in roller ski skating. Int J Sports Med. 2002;23(3):191-5.
13. Minetti AE. Passive tools for enhancing muscle-driven motion and locomotion. J Exp Biol. 2004;207(Pt 8):1265-72.
14. Mittelstadt SW, Hoffmann MD, Watts PB, et al. Lactate response to uphill roller skiing: diagonal stride versus double pole technique. Med Sci Sports Exerc. 1995;27(11):1563-8.
15. Nilsson J, Tveit P, Eikrehagen O. Effects of speed on temporal patterns in classical style and freestyle cross-country skiing. Sports Biomech. 2004;3(1):85-107.
16. Nilsson JE, Holmberg HC, Tveit P, Hallen J. Effects of 20-s and 180-s double poling interval training in cross-country skiers. Eur J Appl Physiol. 2004;92(1-2):121-7.
17. Nummela A, Vuorimaa T, Rusko HK. Changes in force production, blood lactate and EMG activity in the 400-m sprint. J Sports Sci. 1992;10(3):217-28.
18. Nummela AT, Paavolainen LM, Sharwood KA, Lambert MI, Noakes TD, Rusko HK. Neuromuscular factors determining 5 km running performance and running economy in well-trained athletes. Eur J Appl Physiol. 2006;97:1-8.
19. Roy B, Barbeau L. Facteurs d'efficacité dans certaines techniques classiques en ski de fond: le pas de un et la poussée simultanée. STAPS. 1991;12:37-43.
20. Rundell KW, Mc Carthy JR. Effect of kinematic variables on performance in women during a cross-country ski race. Med Sci Sports Exerc. 1996;28(11):1413-7.
21. Rusko H, Nummela A, Mero A. A new method for the evaluation of anaerobic running power in athletes. Eur J Appl Physiol. 1993;66(2):97-101.
22. Smith GA. Biomechanics of cross country skiing. In: Rusko H, editor. Cross Country Skiing: Olympic Handbook of Sports Medicine. Oxford (England): Blackwell Publishing; 2002. p. 32-61.
23. Smith GA, Fewster JB, Braudt SM. Double poling kinematics and performance in cross-country skiing. J Appl Biomech. 1996;12(1):88-103.
24. Smith GA, Heagy BS. Kinematic analysis of skating technique of Olympic skiers in the men's 50 km race. J Appl Biomech. 1994;10(1):79-88.
25. Smith GA, Nelson RC, Feldman A, Rankinen FL. Analysis of V1 skating technique of Olympic cross-country skiers. Int J Sport Biomech. 1989;5(2):185-207.
26. Stöggl T, Lindinger S, Müller E. Analysis of a simulated sprint competition in classical cross country skiing. Scand J Med Sci Sports. 2007;17(4):362-72.
27. Stöggl T, Lindinger S, Müller E. Evaluation of an upper-body strength test for the cross-country skiing sprint. Med Sci Sports Exerc. 2007;39(7):1160-9.
28. Stöggl T, Müller E. Competition analysis of the last decade (1996-2008) in cross-country skiing. In: Müller E, Lindinger S, Stöggl T, editors. 4th International Congress on Skiing and Science (ICSS). St. Anton am Arlberg (Austria): Meyer & Meyer Verlag; 2008. p. 657-77.
29. Stöggl T, Müller E, Lindinger S. Biomechanical comparison of the double-push technique and the conventional skate skiing technique in cross-country sprint skiing. J Sport Sci. 2008;26(11):1225-33.
30. Vandewalle H, Pérès G, Monod H. Standard anaerobic exercise tests. Sports Med. 1987;4(4):268-89.


©2009The American College of Sports Medicine