360 Alex Rosenberg
tinction among explanations.. The distinction is between what are called “how-
possibly explanations” and “why-necessary explanations.” How-possible explana-
tions show how something could have happened, by adducing facts which show
that there is after all no good reason for supposing it could not have happened.
A why-necessary explanation shows that its explanandum had to have happened.
These two different kinds are distinct and independent of one another. Each kind
of explanation will be appropriate to a different inquiry, even when the two differ-
ent inquiries are expressed in the same words.
There is an important asymmetry between how-possible and why-necessary ex-
planations that philosophers of history recognized. Once a how–possible explana-
tion has been given, it makes perfect sense to go on and ask for a why-necessary
explanation. But the reverse is not the case. Once a why-necessary explana-
tion has been given, there is no point asking for a how-possible explanation. For
in showing why something had to happen, we have removed all obstacles to its
possibly happening. Some philosophers of history went on to suggest that why
necessary explanation are “complete”. But this is a notion hard to make clear
in the case of, say, causal explanations, in which it is impossible to describe all
the conditions, positive and negative, individually necessary and jointly sufficient
for the occurrence of an event which we seek to explain. For our purposes all
that will be required is the observation that a why-necessary explanation provides
more information about exactly how its explanans came about than a how-possible
explanation, and that is the source of the asymmetry between them. It is not dif-
ficult to graft this distinction on to the one broached above between erotetic and
pragmatic approaches to explanation. On the erotetic view, whether a question
expresses a request for a how-possible explanation or a why-necessary one is a mat-
ter of the context in which the question is put, the information available to the
interlocutors, their aims and interests. Accordingly, sometimes a why-necessary
explanation will not be an appropriate response to an explanatory question. But
all this is compatible with the fact that a why necessary explanation provides more
information about the causally necessary conditions for the matter to be explained.
The exponent of a non-erotetic approach to explanations will hold that there is
such a thing as a complete and correct explanation independent of contexts of in-
quirer’s questions, and that insofar as they are both incomplete, the how-possible
explanation is more incomplete and the why-necessary closer to the whole story.
The reductionist will sympathize with this view, as we shall now see.
Consider the ultimate explanation for eyespots in the buckeye butterfly species
Precis coenia. Notice to begin with there is no scope for explaining the law that
these butterflies have eye-spots, or patterns that may include eye spots, scalloped
color patterns, or edge-bands, even though almost all of them do have such mark-
ings. There is no such law to be explained, as there are no laws about butterflies,
still less any species of them. That the buckeye butterfly has such eyespots is
however a historical fact to be explained.
The ultimate explanation has it that eyespots on butterfly and moth wings have
been selected for over a long course of evolutionary history. On some butterflies