Martin (1992: 538):
Barthes continues by pointing out that activity sequences have names, and may be encapsulated by the name to form part of another sequence. This suggests in effect that the notion of constituency can be applied to the compositional relations among activity sequences (and Barthes in fact provides an appropriate tree diagram for the first episode of Goldfinger):
It (a sequence) is also founded a maximo: enclosed on its function, subsumed under a name, the sequence itself constitutes a new unit, ready to function as a single term in another, more extensive sequence. Here, for example, is a micro-sequence: hand held out, hand shaken, hand released. This Greeting then becomes a simple function: on the one hand, it assumes the rôle of an indice (flabbiness of Du Pont, Bond's distaste); on the other, it forms globally a term in a larger sequence, with the name Meeting, whose other terms (approach, halt, interpellation, sitting down) can themselves be micro-sequences. (1977: 102-3)
The meeting sequence in other words has as one of its nuclei another sequence — greeting:
meeting (greeting): approach ^ halt ^ interpellation ^ (hand held out ^ hand shaken ^ hand released) ^ sitting down
 For Barthes, it is a sequence of language that has a name. Martin's activity sequence is misconstrued as context (field) which, in turn, is misconstrued as register.
 The (tautological) claim here is that part-whole relations (constituency) can be applied to part-whole (compositional) relations.
 To be clear, Barthes claim is that (micro-)sequences can be embedded in other sequences.