It can be argued (cf. Dizadji‑Bahmani et al. 2010) that an increase in
coherence is one goal that drives reductionist enterprises. Consequently, the
question if or how well this goal is achieved can serve as an epistemic criterion
for evaluating both a concrete case of a purported reduction and our model of
reduction : what conditions on the model allow for an increase in coherence ?
In order to answer this question, I provide an analysis of the relation between
the reduction and the coherence of two theories. The underlying model of
reduction is a (generalised) Nagelian model (cf. Nagel 1970, Schaffner 1974,
Dizadji‑Bahmani et al. 2010). For coherence, different measures have been put
forward (e.g. Shogenji 1999, Olsson 2002, Fitelson 2003, Bovens & Hartmann
2003). However, since there are counterexamples to each proposed coherence
measure, we should be careful that the analysis be sufficiently stable (in a sense
to be specified). It will turn out that this can be done.