Thus far, modern science has relied on an essentially analytic strategy. This strategy has proved immensely successful, at least for some regions of reality. Other regions have proved more refractory, because there are items that cannot be fragmented without losing relevant information. The misplaced faith in analytic, reductionist methods follows almost unavoidably from the lack of a robust ontology, and in particular from the lack of both a theory of wholes and their parts and a theory of levels of reality. The paper unfolds the multi-layered complexity of the connections linking together analysis and synthesis and exhibits a few landmarks which should be exploited further, such as the ontological relevance of windowing; the typology of wholes, including their relationships and in particular the whole-whole family of connections; and the theory of levels of reality, or what can be termed an ontological theory of types.