is an analogous feature, but the explanation of why parthood is analogous will leave
open whether the more natural (and more topic-restricted) parthood relations are
each as natural as each other, with none of them being the“central kind”of parthood.
Sometimes this unity of analogy is via relations to a single specification. But
I believe in other cases it is not, but rather it is simply a fact that a set of specifications
have a kind of internal unity, even though we cannot explicate in what this unity
consists.^8 In those cases, I can’t give a criterion for when a feature is an analogous
feature as opposed to a merely disjunctive feature. It might be that in this sort of
case there is little more that we can do than take the difference between analogous
and disjunctive properties as a brute difference in how natural the properties in
question are.
The following remarks give barely more than the appearance of precision, but
might still be of some use. Consider two functions on sets of properties. Thefirst
function takes a set of properties to the mere disjunction of the members of that set.
The second function takes a set of properties to an analogous property“derived
from”the members of that set. Mere disjunctions and analogous properties are
always less natural than their disjuncts or analogue instances. But disjunctive prop-
erties are far less natural than their disjuncts, whereas analogous properties can be
almost as natural as their analogue instances. This gives some content to the idea that
some properties are“unified by analogy”whereas others are unified by nothing more
than a mere list of the actual or possible things that have them.
This way of talking suggests that properties can be necessarily coextensive yet non-
identical. I am inclined to think that properties are individuated morefinely than
necessary coextension, but a more cautious statement is this: some“disjunctive”
properties are less“disjunctive”than others. There are equinumerous sets of prop-
erties H and P such that there is a 1–1 correspondence between H and P that
preserves degree of naturalness, and yet the analogous property consisting of
the members of H is more natural than the disjunctive property consisting of the
members of P. And even this way of speaking might be more committal than we
want, since it requiresdegreesof naturalness. An even more cautious statement,
which I owe to Joshua Spencer, is this: there are two sets of properties, H and P, and a
1 – 1 correspondence,f, between the members of H and the members of P such that
for eachxof H,xis as natural asf(x)and yet the“disjunction”of the properties in
H is more natural than the disjunction of the properties in P.
My conception of an analogous property is in many ways less sophisticated than
the conceptions of analogy of my medieval predecessors from whom I have appro-
priated the term. There are many ways in which the analogue instances of an
analogous property could be related to each other in such a way that the analogous
property enjoys more unity—that is, a higher degree of naturalness—than a mere
(^8) Pasnau and Shields (2004: 116–19) discuss what they call“analogical predication,”a species of which
is ordered analogical predication. This suggests that they recognize other species of analogical predication.