The biomechanical characteristics of the strongman atlas stone lift

Experimental approach

A cross-sectional observational experimental design was used to describe the biomechanical characteristics of athletes performing the atlas stone lift and assess temporal and kinematic measures of an incremental mass, four atlas stone series. Well trained strongman athletes with strongman competition experience (Table 1) undertook two testing sessions. Session one consisted of a 1RM atlas stone lift to establish loading conditions for session two. Session two consisted of the collection of temporal and kinematic measures during three sets of four lifts of atlas stones of incremental mass (up to ~85% 1RM) over a fixed-height bar. Body mass, trochanterion-tibiale laterale height and tibiale laterale height anthropometric measures were taken by a trained person using ISAK methodologies (Marfell-Jones, Stewart & De Ridder, 2012) to assist in describing the study population.

Participants

Twenty experienced strongman competitors (12 male and eight female) were recruited from two local strongman gyms (Table 1). All participants were required to have a minimum of 18 months strongman training experience, have competed in a minimum of one strongman competition and be free from moderate or major injury for at least one week prior to testing. A moderate injury was defined as an injury that had stopped the athlete from performing a particular strongman exercise during a strongman session, while a major injury was defined as an injury which prevented the athlete from continuing with all exercises and/or the session completely (Winwood et al., 2014b; Keogh & Winwood, 2017). Participants meeting the above criteria were informed of the purpose of the study and asked to sign an informed consent form. Ethical approval was granted for all procedures used throughout this study by Bond University’s Human Research Ethics Committee (BH00045).

Trial conditions

To achieve optimal performance during the session, athletes were asked to prepare for each session in the same way in which they would prepare for a regular training session. Due to the range of individual loading parameters and experience level of all athletes recruited in the study, self-directed warm up routines were performed by each athlete (Winwood et al., 2014a, 2015a, 2015b; Renals et al., 2018; Winwood et al., 2019). Warm up routines lasted ~15–30 min and included repetitions of the atlas stone lift at loads approaching those expected to be used by the individual throughout the session. Generally, athletes would begin their warm up with dynamic stretching, including resistance band exercises, followed by barbell-only (no additional load) squats or deadlifts. Athletes would move on to stone pickups (either performing a Romanian deadlift-like pickup of the stone from the ground, or lifting the stone in a full range of motion to bar height without passing the stone over the bar) at low loading (~<60% 1RM). As athletes approached stone masses expected to be used in the session, the athlete would begin to complete full stone lift repetitions where the stone was passed over the bar. Athletes were permitted to use knee and elbow sleeves, lifting belts, arm/wrist wraps and tacky during sessions, as this is standard equipment used in competition and training.

Session protocols

Session one 1RM testing required athletes to lift a stone of greatest mass over a bar of fixed height (female: 1.2 m; male: 1.3 m). Athletes worked up to their heaviest stone in mass increments selected by the athlete. Mass increments were dependent on the mass of the stones available, the perceived effort of the previous lift and current training loads used by each participant. When an athlete failed to lift the stone over the prescribed height bar, the athlete was given one additional attempt to successfully complete the lift. Athletes were assigned rest periods of six to eight minutes between each stone attempt (Winwood, Keogh & Harris, 2011). The mass of the heaviest stone the athlete was able to successfully pass over the bar was determined to be their 1RM.

Session two was performed a minimum of seven days after session one and required athletes to perform three sets of a four stone series over a bar (female: 1.2 m; male: 1.3 m) as quickly as possible. Each stone within the series were of incremental mass, where stone one (repetition one) ≈60% 1RM, stone two (repetition two) ≈70% 1RM, stone three (repetition three) ≈80% 1RM and stone four (repetition four) ≈ 85% 1RM (Table 2). As is the nature of the atlas stone, stones were of a fixed mass (mass could not be added or removed from the stone), therefore stones within each series were selected based on the closest stone mass available to fit the required percentage of 1RM for each participant. The diameter and surface finish of stone varied with the mass of the stone (Table 2).

To begin each set, the athletes were positioned in the typical atlas stone competition starting position with the stone on the ground between their legs and their hands resting on the bar for which the stone was to be passed over. On the signal “athlete ready, three, two, one, lift” the participant commenced lifting stone one over the bar. After the completion of each repetition, the next stone in the series was positioned in front of the participant by a trained loading assistant. When an athlete was unable to pass a stone over the bar or the final stone in the series was successfully passed over the bar the trial was concluded, with each series typically completed in 60 s.

Data acquisition and analysis

Methodologies of Hindle, Keogh & Lorimer (2020) were used to estimate joint kinematics of athletes performing the atlas stone lift. Four magnetic, angular rate and gravity (MARG) devices (ImeasureU, Vicon Motion Systems Ltd., Oxford, UK) were used to capture acceleration, angular velocity (1,125 Hz) and magnetic field strength data (112 Hz). MARG devices were positioned on the pelvis (halfway between the left and right posterior superior iliac spine), right thigh (approximately 150 mm proximal to the lateral epicondyle of the femur), right shank (approximately 100 mm distal to the lateral tibial condyle) and right foot (midway between the base of the foot and the lateral malleoli) (Hindle, Keogh & Lorimer, 2020). The MARG data collected for each segment were input into a custom Matlab script (The Mathworks Inc., Natick, MA, USA) to measure hip, knee and ankle joint angles in the sagittal plane (Hindle, Keogh & Lorimer, 2020). The methodology has shown acceptable to excellent agreement with optical motion capture methodologies in similar movements such as the squat, box squat and sandbag pickup (Hindle, Keogh & Lorimer, 2020).

Two video cameras (iPad Air 2, iOS 13.3.1, Apple Inc., CA, USA) were used to capture video data at 30 Hz (Fig. 2). Video data were synchronised with MARG data using the ground impact of a submaximal jump performed immediately prior to the commencement of each set. The video data allowed for the calculation of the temporal parameters (phase duration, repetition duration), while joint kinematics at various instances throughout a repetition were obtained from the time-synched MARG data. Temporal and kinematic measurements assessed during each repetition of the atlas stone lift are defined in Table 3, with a pictorial representation of each phase of the lift presented in Fig. 3. Joint angle conventions are outlined in Fig. 4.

Statistical methods

Data were checked for normality using visual inspection and a Shapiro Wilks test. Homogeneity of variances were checked using Levene’s test, homogeneity of covariances were checked using Box’s M-test (p < 0.001) and sphericity was checked throughout the computation of ANOVA tests. Mean and standard deviations of all variables were calculated for all phases throughout the stone lift. The joint kinematic results for the recovery and initial grip phases were not presented due to the high variability in the participants’ movements observed in these non-lifting, preparation phases, thus statistical analyses of these phases were not performed. A one-way repeated measures ANOVA test was used to establish the biomechanical characteristics of the lift by comparing: (1) between phase characteristics; (2) between repetition characteristics; and (3) between set characteristics. Between set statistical analysis was performed prior to further analyses to assess if data from each of the three sets could be combined. A two-way mixed model ANOVA test was used to identify interactions of sex and repetitions for each biomechanical characteristic. Partial eta-squared effect sizes (η_p²) were calculated for two-way interactions with classifications of negligible (η_p² ≤ 0.01), small (0.01 > η_p² ≥ 0.06), medium (0.06 > η_p² ≥ 0.14) and large (η_p² > 0.14) (Cohen, 1988). Bonferroni post-hoc pairwise t-tests were conducted on parameters where significant differences were detected. Cohen’s d (d) effect sizes were calculated for t-tests with classification of negligible (d < 0.2), small (0.2 ≤ d < 0.5), medium (0.5 ≤ d < 0.8) and large (d ≥ 0.8) (Cohen, 1988). Post-hoc intra-class correlation coefficient (ICC) and standard error of measurement (SEM) metrics were calculated to assess relative and absolute reliability of each biomechanical measure, respectively. Reliability was classified as poor (ICC < 0.5), moderate (0.5 ≤ ICC < 0.75), good (0.75 ≤ ICC < 0.9) and excellent (ICC ≥ 0.9) (Koo & Li, 2016). Statistical analyses were performed in R version 3.6.1 (R Development Core Team, Vienna, Austria), with statistical significance accepted at p < 0.05 unless otherwise stated.

General biomechanical characterisation—sex independent

The atlas stone lift could be characterised by: maximal hip and moderate knee flexion and ankle dorsiflexion at the beginning of the first pull and maximal hip ROM throughout the first pull; moderate hip and knee flexion and moderate ankle plantarflexion at the beginning of the lap phase and minimal hip, knee and ankle ROM throughout the lap phase; moderate hip and maximal knee flexion and ankle dorsiflexion at the beginning of the second pull phase and maximal knee and ankle ROM throughout the second pull phase; and maximal hip and knee extension and ankle plantarflexion at lift completion (Fig. 5, Tables S1, S2, S3).

Excluding the recovery and initial grip phases, the second pull phase was statistically longer in duration than all other lifting phases (0.27 ≤ d ≤ 1.12, p < 0.001), followed by the lap phase which was statistically longer in duration than the first pull phase (d = 0.34, p < 0.001) (Fig. 5, Table S3).

General biomechanical characterisation—sex dependent

When compared with male athletes, female athletes exhibited: greater hip flexion and ankle plantarflexion at the beginning of the first pull (0.78 ≤ d ≤ 1.21, p < 0.001) and greater overall hip ROM throughout the first pull (d = 0.56, p < 0.001); greater hip flexion and knee extension at the beginning of the lap phase (0.58 ≤ d ≤ 0.77, p < 0.001), and smaller hip and ankle ROM throughout the lap phase (0.26 ≤ d ≤ 0.46, p ≤ 0.049); greater hip flexion, knee extension and ankle plantarflexion at the beginning of the second pull phase (0.29 ≤ d ≤ 0.48, p ≤ 0.034), and smaller knee ROM and greater ankle ROM throughout the second pull phase (-0.53 ≤ d ≤ 0.32, p ≤ 0.021); and greater hip flexion and ankle plantarflexion at lift completion (0.41 ≤ d ≤ 0.85, p ≤ 0.003) (Fig. 5, Tables S1, S4).

Few statistical between-sex temporal differences were observed (Table S5). Male athletes displayed a statistically longer second pull phase duration than female athletes (d = 0.42, p = 0.012) (Fig. 5, Tables S1, S4).

Between repetition biomechanical differences—sex independent (main effect)

Statistically significant between-repetition differences were most commonly observed for joint kinematics between combinations of the first two repetitions and the last two repetitions of the set (e.g., between repetition one-two and three-four) (Figs. 6, 7, 8, 9, Table S5). First pull phase hip and ankle ROM was smaller in repetition one than the final three repetitions (−0.717 ≤ d ≤ −0.496, p ≤ 0.002) (excluding repetition two ankle ROM). Lap phase hip and ankle ROM was smaller in repetition one than the final three repetitions (−1.15 ≤ d ≤ −0.46, p < 0.001), and smaller in repetitions two and three (hip only) than repetition four (−0.65 ≤ d ≤ −0.37, p ≤ 0.003). No statistical between-repetition differences were observed at any joint for the position in which athletes began the second pull phase (Table S5).

For each repetition, individual phase durations and total repetition duration increased as the set progressed (Fig. 10, Table S6), with medium to large effect sizes recorded between repetition one and repetitions three and four (0.64 ≤ d ≤ 1.73, p ≤ 0.003). Where statistical differences were reported for phase duration between sequential stones (e.g., repetition one vs repetition two, repetition three vs repetition four), smaller effect sizes were typically observed (0.31 ≤ d ≤ 1.03, p ≤ 0.005) (Table S6).

Between repetition biomechanical differences—sex dependent (two-way interaction)

While not evident in female athletes, male athletes generally displayed: smaller hip ROM during the second pull phase of the first three repetitions when compared with the final repetition (−0.87 ≤ d ≤ −0.59, p ≤ 0.011); smaller hip extension at lift completion during the first two repetitions of the set when compared with the final two repetitions (−1.24 ≤ d ≤ −0.55, p < 0.038); and greater plantarflexion of the ankle at lift completion in the first repetition when compared with the final repetition (d = 0.75, p = 0.014) (Tables S5, S6, S1). No temporal two-way interactions between sex and repetition were observed (Table S5).

Between set biomechanical differences

Between-set analysis was performed so to identifying any potential effects of set number on the biomechanics of the athlete. Hip flexion was greater at the beginning of the first pull, lap phase and second pull in set one than set two and three (0.04 ≤ d ≤ 0.26, p ≤ 0.013) (Tables S7, S8, S9). Second pull duration was significantly greater during set one than set three (d = 0.19, p = 0.012) (Tables S8, S9). No statistical between-set difference in total repetition duration was observed for any repetition.

General biomechanical characterisation—sex independent

To describe the general movement pattern of the atlas stone lift, hypothesis one sought to determine if the various phases of the atlas stone lift were biomechanically similar to selected traditional weight training exercises.

General biomechanical characterisation—sex dependent

A number of between-sex differences in joint kinematics were observed. Most notably, female athletes exhibited greater hip flexion (female: 84.7 ± 18.7°; male: 63.7 ± 15.8°) and ankle plantarflexion (female: 0.3 ± 8.4°; male: 6.0 ± 6.0°) at the beginning of the first pull, lap and second pull phase than male athletes.

The between-sex difference in hip flexion at the beginning of the first pull may be the result of the differences in anthropometric ratios of the female and male population. At the beginning of the first pull, a greater arm to lower limb length ratio would enable an athlete to grip the bottom of the stone with less flexion of the hip (assuming constant knee flexion angle). Keogh et al. (2008) reported statistically greater arm to leg length ratios in male powerlifters (67.8% ± 2.9%, n = 54) when compared with female powerlifters (64.5% ± 2.5%, n = 14), supporting the deduction that the between-sex differences observed in hip flexion at lift off for the atlas stone lift may be partially due to the anthropometric differences between male and female strength athletes.

The smaller hip flexion displayed by male athletes at the beginning of the lap and second pull phase may be a mechanism used by male athletes to accommodate the larger diameter stone (typically lifted by male athletes when compared with female athletes) so to ensure the COM of the stone remains as close as possible to their COM and within their base of support. The compensative mechanism of greater hip extension may result in a similar stone to body COM distance and thus resistive moment arm length about the lumbar spine in male and female athletes. Although not measurable in the current study, such a result has been reported in a study in which males had significantly greater absolute but not relative L5/S1 joint moments than females when lifting boxes between 15–24 kg from a pallet at a self-selected pace (Plamondon et al., 2014). The between-sex differences in hip, knee and ankle joint kinematics and phase duration measures observed while athletes performed the atlas stone lift are in support of hypothesis two.

Between repetition biomechanical differences—sex independent (main effect)

Hip and ankle joint ROM during the initial pull and lap phase of the lift were generally smaller for athletes during repetition one when compared with the final three repetitions. Greater flexion of the knee and hip at the beginning of the first pull were generally observed in the first two repetitions when compared with the final two repetitions.

The smaller hip and ankle ROM in the initial repetitions than the later repetitions indicate athletes performed abbreviated versions of the lift to begin the set. The strategy of athletes performing an abbreviated version of the lift is likely executed with the intention of self-preservation of energy (Hooper et al., 2014) and conservation of overall repetition and set time. This is supported by the statistically shorter phase durations and total repetition duration observed during the first two repetitions when compared with the final two repetitions of the set. The increased hip ROM when lifting the greater mass stones is also in line with previous research on load-dependant biomechanical differences observed during the back squat (Yavuz & Erdag, 2017).

Although fatigue was not directly measured in this research, the very short recovery duration between each repetition may contribute to some level of athlete fatigue. Recovery phase duration was found to increase as athletes progressed through the set of four atlas stone lift repetitions. Where the onset of fatigue is observed, research has demonstrated significant changes in joint kinematics of male participants performing a box lifting task (Trafimow et al., 1993). Such previous research may suggest that some of the between repetition differences observed in the current study be due to the acute effect of fatigue that progressively increased within the set of incremental mass stone lifts. In support of hypothesis three, a number of between-repetition differences were observed in athletes performing the atlas stone lift. Further, a large portion of between-repetition differences observed were between repetition one and four.

Between repetition biomechanical differences—sex dependent (two-way interaction)

Male athletes exhibited smaller hip ROM during the second pull phase of the first three repetitions when compared with the final repetition and smaller hip extension at lift completion during the first two repetitions of the set when compared with the final two repetitions. Female athletes appeared to use a more consistent technique throughout the four repetitions, whereby they did not exhibit these significant between repetition differences.

To ensure the bottom of the stone cleared the height of the bar in the final two repetitions, male athletes appeared to use greater extension (often hyperextension) of the hip. The greater extension of the hip at lift completion, likely contributed to the greater hip ROM displayed by male athletes in the final repetition when compared to the first three repetitions.

While the two-way interactions between sex and repetition further support hypothesis three, the exact reasoning behind the different mechanisms used throughout the set by male and females is somewhat unclear. Future researchers may look to investigate how between-sex differences in anthropometry, motor control and muscle recruitment strategies contribute to the kinematic between-sex differences observed during the atlas stone lift series.

Additional considerations

The current study is not exempt from limitations. As with any research, care should be taken when interpreting comparative results between groups, ensuring the magnitude of the error of the measurement system is recognised. In the case of the temporal parameters, the measurement accuracy was limited by the frame rate of the video camera, while kinematic parameters were limited by the accuracy of the MARG-based motion capture methodology (Hindle, Keogh & Lorimer, 2020). Good (ICC ≥ 0.75) to excellent (ICC ≥ 0.9) relative reliability was generally found for all biomechanical parameters measured within the study using the MARG and video camera methods (Table S10).

Twenty experienced strongman athletes (12 male, eight female) were recruited for the study. While the combined number of male and female strongman athletes recruited in the current study is much larger than the number of strongman athletes recruited in any previous strongman exercise biomechanics study, the individual number of male (n = 12) and female (n = 8) participants is similar or only slightly larger than previous research (McGill, McDermott & Fenwick, 2009; Keogh et al., 2010a; Keogh et al., 2010b; Keogh et al., 2014; Winwood et al., 2014a, 2015a, 2015b; Renals et al., 2018). A greater number of both male and female athletes would strengthen the conclusions drawn from the observed between-sex biomechanical differences.

Variation in the increments of the mass of the stones, dimensions of stones and surface finish of stones may also be viewed as a limitation to this study. Variable increments, dimensions and surfaces of stones, is however a reality of the sport of strongman and provides greater insight into the realities of strongman biomechanics.

As this is the first biomechanics study to describe kinematic and temporal parameters of athletes performing the atlas stone lift there is much scope for future research, including: transverse and frontal plane joint kinematic analyses; establishing relationships between anthropometrics of strongman athletes and their biomechanical characteristics; the effect of stone dimension, mass and surface finish on the biomechanics of an athlete; the injury risks associated with the atlas stone lift; and the biomechanical determinants of greater performance in the atlas stone competition event.