The fatigue protocol consisted of intermittent isometric contractions at 50 % MVC, a level at which nearly all the motor units of the FDI are activated (Milner-Brown et al. 1973; De Luca et al. 1996). To determine the maximal voluntary torque produced by the muscle, each experimental session began with the subject performing three MVCs of 4-5 s duration separated by 2-3 min pauses. The motor units were subsequently identified by means of a template-matching algorithm program (see below) from records obtained during two or three isometric ramp contractions. Once motor unit action potential(s) had been clearly identified (usually 1-3), the subject maintained for 0.5-1 min a level of torque for each motor unit that was just sufficient to sustain the discharge frequency at a constant low rate (7-10 Hz). In the high-threshold motor units, the contraction was interrupted by one or two small pauses to minimize fatigue. These data were collected for the purpose of mechanical recording of single motor units by the spike-triggered averaging.
After the prefatigue recordings, the fatigue task was performed. It consisted of the subjects following a target on the screen of an oscilloscope and reaching 50 % MVC in 3 s, maintaining this level for 10 s, and returning slowly to the baseline in 3 s. A rest period of 4 s was allowed between two successive contractions. This protocol, which was very similar to the one proposed by Enoka et al. (1989), was repeated 3 times min−1 until the subject was unable to maintain the required level of force for three successive contractions (the endurance limit). When a motor unit had been successfully tracked throughout the fatigue test, its discharge during sustained threshold contraction was again collected using the same procedure and for the same purpose as in the prefatigue condition. For some motor units, the recovery from fatigue was tested 15 and 30 min after the fatigue test by recording its discharge during sustained threshold contraction and recruitment threshold during ramp contractions. In other experiments, the muscle was stretched by maximal adduction of the index finger for 15 s after the post-fatigue recordings, and the discharge during sustained threshold contraction and recruitment threshold of the motor unit were again recorded.
Data processing was performed off-line from taped records (Vetter 620, Rebersburg, PA, USA or Sony PCM-DAT, DTR 800, Bio-Logic, Claix, France). The signals recorded during the fatigue tests were digitized at a frequency of 10 kHz and analysed with AcqKnowledge 3.2.4 software (BIOPAC Systems Inc., Santa Barabara, CA, USA). The recruitment and derecruitment thresholds were determined during each ramp contraction for the selected motor units. The recruitment threshold was defined as the torque at the which the motor unit began to fire during the increasing phase of the contraction, whereas the derecruitment threshold was defined as the torque level corresponding to the last pulse of this unit during the decreasing phase of the contraction. In addition, the mean discharge frequency of each unit was measured during the force plateau and only units firing continuously with a minimum of 20 pulses were analysed. Motor unit discrimination was accomplished using a computer-based, template-matching algorithm (Signal Processing systems, SPS 8701, Malvern, Victoria, Australia). Action potentials of single motor units were identified on the basis of amplitude and waveform shape. Only the motor units that were clearly identifiable for the entire course of the fatigue test were included in the analysis. The mechanical recording of a single motor unit was performed by spike-triggered averaging with a digital oscilloscope (model 4094c, Nicolet, Madison, WI, USA, sampling rate: 20 kHz). Usually, 50-200 sweeps were averaged, depending upon motor unit size. For large units (recruitment threshold > 25 % MVC), 50 sweeps were often sufficient. The peak amplitude, the time to peak, and the time to half-relaxation of the mechanical response were measured for each unit.
The significance of the changes in the different parameters during fatigue and recovery was assessed either using Student's paired t test or repeated-measures analysis of variance (ANOVA). When significant main effects were found with an ANOVA, the Newman-Keuls test was used to locate the significant differences between the means. In the text, values are given as means ±s.e.m. and a significance level of P < 0.05 was used for all statistical comparisons.