Analysis of simultaneous measurements of the channel-base current and the vertical electric field 30 m from triggered lightning reveals that the fields associated with M components, although essentially electrostatic, appear to be proportional to the time derivatives of the associated M currents. Based on this finding, coupled with other observations and modeling, a mechanism for the lightning M component is proposed. According to this mechanism an M component involves a downward progressing incident wave (the analog of a leader) followed by an upward progressing reflected wave (the analog of a return stroke). However, as opposed to a leader-return stroke sequence in which the latter removes the charge deposited by the former, both the upward and the downward processes contribute about equally to the total charge flowing from the bottom of the channel at any instant of time. Such a mode of charge transfer to ground, distinctly different from a leader-return stroke sequence, is possible because of the presence of a path capable of supporting the propagation of a traveling wave (facilitated by a continuing current flowing to ground) and the fact that the ground is essentially a short circuit for the downward incident wave, so that the magnitude of the current reflection coefficient at ground is virtually equal to unity. We show that some observed properties of M components can be explained if the lightning channel traversed by an M-current wave is represented as a linear R-C transmission line. In this view, the preferential attenuation of the higher-frequency components on an R-C line is responsible for the lack of frequencies above several kilohertz in both the M-current pulses measured at the channel base and the M-light pulses observed in the bottom 1 km or so of the channel. Further, the relatively high characteristic impedance of the channel, of the order of tens to hundreds of kilohms for frequencies below some kilohertz, inferred from the linear R-C line approximation, is consistent with the observation that even a relatively poor ground is sensed by an incident M wave as essentially a short circuit. However, on a linear R-C transmission line the higher-frequency components travel faster than lower-frequency components (this velocity dispersion implying that the original pulse would spread while propagating along the line), whereas the shape of the M-light pulses does not change much within the bottom 1 km or so, as if the channel were a distortionless transmission line. We speculate, on physical grounds, that the front of the traveling M-current pulse heats the channel so that the pulse tail encounters a lowered resistance and, as a result, accelerates. By virtue of these two opposing effects, velocity dispersion and channel nonlinearity, an M pulse is formed whose more-or-less symmetrical shape is preserved over a relatively large distance, as in the case of a soliton.