Thanks Armin, Simon and all. It is good to have this resolved by the WG, and I note also the various qualifying comments and concerns in the discussions on the WG list. As Simon points out, Schema.org comes with its own heritage/legacy of terms and structures. For example we also already have two mechanisms for describing quantities (see
https://github.com/schemaorg/schemaorg/issues/1390#issuecomment-252202548 and nearby). There is also interest in figuring out a story for how high-level descriptions of datasets relate to their actual data contents (cf. W3C CSVW and also W3C Data Cube, which has some bearing on the observation/measurement topic). All of which is just to say that whatever we do in Schema.org may not be entirely 1:1 with SOSA, but we shall try to avoid needless divergence, and to document (machine readably when possible) the commonalities and differences as things mature. And of course we'll credit the WG appropriately for any Schema.org terms that are based on this WG's efforts.