We applied Fast Fourier Transform to monopolarily recorded surface myoelectric signals from m. biceps brachii to obtain the turning frequency (TF) at which the signals divide into high and low frequency components. Then using the TF as a cut-off point, the raw myoelectric signals were divided into high frequency component (HFC) and low frequency component (LFC) .
1) TF values were constant at every recording position along muscle fibers. And there was no considerable relationship between developed tensions and TF values.
2) Cross-correlation analysis was done to obtain the phase relationship between LFCs, and between HFCs at four different recording positions along muscle fibers. LFCs appeared in phase in all combinations below the tension of 10%MVC and also in HFCs. Above 20%MVC, LFCs represented the time delay depending upon the electrode distance in six subjects. The conduction velocity calculated from the relationship between time delay and distance in LFC was too high (16.1 m·s^-1-33.3 m·s^-1) compared with the muscle fiber and/or motor unit conduction velocity. But LFCs in other subjects remained in phase. HFCs showed no considerable relationship above 20%MVC in six subjects, while other subjects remained in phase between HFCs.
The present results differ from the recent investigation in m. vastus medialis (Kamo & Morimoto, 2000) . Our proposal on the construction mechanism of surface myoelectric signals could not be adapted to the electrical signals from m.biceps brachii. The present results suggest the different in architecture of muscle fibers and the innervation of the motor nerve in a motor unit between two muscles.

In this study, we investigated the applicability of the analyzing method of surface myoelectric signal proposed by Kamo & Morimoto (2000) for children's (age groups: Grade 2, 4, and 6 of elementary school) monopolarily recorded surface mvoelectric signal from m. vastus medialis. The subjects (n=16) were requested to exert constant brief (10 sec) isometric knee extension at low tension (4-12%MVC) .
Next, to elucidate the age dependent change of the possible motor control mechanisms, the change of the contribution of the integrated value of low (LFC) and high frequency component (HFC) to integrated value of raw signal was tested. Index of the contribution was considered from the slope of the regression line between raw and LFC integrated values, and between raw and HFC integrated values.
Obtained results were as follows:
1. Turing frequency (TF) appeared in the children's amplitude spectrum of surface mvoelectric signals. And TF shifted toward higher frequency depending upon the developed tension.
2. For children, cross-correlation analysis characterized LFC as conducting, and HFC as non-conducting characteristics.
3. There was no significant difference in integrated value of raw signal between age groups at each target tension. Results were similar concerning LFC and HFC as well.
4. The comparison of regression slopes showed that HFC had significantly lower contribution to construct the raw signal than LFC. Furthermore, results showed no age related difference as far as the contribution of the LFC, HFC to the raw signal concern.
Results 1 and 2 were in good agreement with the results obtainend from university-students (n=5) . Therefore, it suggests that the analyzing method proposed by Kamo & Morimoto (2000) . can adapt to the mvoelectric signal from children.
Similar result between the age groups suggests that under low-tension brief contraction motor unit control mechanism showed no developmental stage difference from 2 nd grade to university student, yet muscular strength of the 2 nd grade did not reach to the university student level.