The importance of semi-direct processes in E1 and E2 capture reactions has been well documented. Little is known, however, about M 1 capture although a vast body of data indicates a non-statistical nature. We discuss here the importance of the semi-direct M 1 process and concentrate in particular on recent experimental results claiming that the M 1 strength in 29Si is nearly one order of magnitude smaller than in 2ssi, the target nucleus. We use shell model arguments to demonstrate that this reduction is indeed consistent with the semi-direct reaction process and that other mechanisms like 2p-2h fragmentation and phonon coupling are not likely to be instrumental in elucidating further the non-statistical nature of the 2aSi 4-n reaction.PACS: 25.40.Lw capture. This was recently exploited by Castel and Mahaux [2] to demonstrate that the success of the valence model in explaining E1 capture in 28Si is in fact due to the dominance of the external contributions to the valence capture integral. This work is extended here to the study of M 1 capture in order to answer some important questions raised by recent experimental investigations. We concentrate in particular on the recent claim by Weigmann et al. that the M 1 strength in 29Si following s-wave neutron capture is surprisingly smaller by nearly one order of magnitude compared to the M 1 strength in 28Si, the target nucleus. One of the aims of this paper will be to point out that this reduction is indeed consistent with an analytical application of the semi-direct formalism and that other mechanisms like 2p-2h fragmentation and phonon coupling are not likely to contribute to the nonstatistical nature of the 2sSi 4-n reaction.