Objectivists who hold that colours have hidden objective essences customarily draw upon standard examples within science, where microstructural properties are called upon to explain why objects have certain `surface' properties, e.g., solidity, solubility, temperature, elasticity, fitness, refractive power, and so on. Such examples are meant to provide models for thinking about colours, in that they offer cases in which the surface property is reduced to, or identified with, a certain microstructural property, e.g., where light is taken to be a form of electromagnetic radiation, temperature of a gas mean kinetic energy of the constituent molecules, and so on.

The logic of scientific identifications or reductions, however, is not as clear as it might be. The concept of temperature, for example, had a distinguished scientific status long before the atomic/molecular theory emerged. If temperature originally was conceptualised as `that property which occupied such and such a causal role' then the way was left open to identify temperature with mean kinetic energy. It is not at all clear, however, that this was the original scientific concept of temperature. A different way of viewing the situation is to take it that the scientific reduction (or identification) of temperature to (with) kinetic energy worked as a two stage process. It first involves a replacement of the original concept by a reconstructed one, followed by a second stage in which the reconstructed concept allows for the identification of temperature with the appropriate property described within statistical mechanics.

However not every case in science in which we can explain surface properties in terms of microstructural processes and properties are ones that involve identification or reduction. Take the example of solidity. It would seem that we have discovered what the microstructural property is that provides the causal basis for solidity. It is not at all clear, however, that solidity has been reduced to or identified with, that microstructure. It all depends on what exactly the concept of solidity is and there are different models for thinking about this concept. Not all of them lend themselves to identification and reduction, and those that do not have strong claims for legitimacy.

To appreciate such models, let us consider what reduction or identification in the case of solidity requires. To explain how solidity has been reduced, we need to specify what the original concept of solidity was. There is, however, more than one candidate. Each candidate is associated with certain causal powers: relative impenetrability, stability of a certain kind, capacities to resist, and so on. There are however, different ways in which solidity might be related to these causal capacities:

1.solidity = the causal capacities and powers to . . .
2.solidity = that microstructural basis whatever it is, that is the causal basis for the causal capacities and powers, as in 1.
3.solidity = the property of having some intrinsic structure whereby the object has the capacities as in 1.

 We need to distinguish account 2 from a cousin:

2*. solidity = that microstructural basis whatever it is, which, as it happens, is the causal basis for the causal capacities and powers, as in 1.

The difference between Models 2 and 2* rests on how the relevant causal powers are related to the microstructure. The term "solidity" is understood on either model as functioning as a name for the property but they are different types of name. In model 2 the characterisation in terms of the causal powers is essential to the understanding of the name, whereas in Model 2* it is not. In the latter case, the causal capacities are used to refer to the microstructure, but any of a number of other characterisations might have served. The causal capacities might not even hold of that particular microstructure, and yet it could refer all the same. Model 2* seems to fit names such as `gold' and `water' _ at least as far as capacities such as appearances and tastes are concerned. Model 2 would seem to fit terms such as `electron', `proton', `gravity wave', `force', etc., and it seems more appropriate than the other for `solidity' and `liquidity'.

On model 2, solidity can be identified with some microstructural property. Reference to the relevant causal capacities is essential to the conceptualisation of solidity but the causal capacities are not essential to solidity itself. On the other hand, with respect to Models 1 and 3, the causal powers are essential to solidity. Given that solidity is conceptualised in these ways the causal powers are essential to the property of solidity and not just to the conceptualisation.

Whichever model we adopt, we can agree that solidity has been explained by reference to certain microstructures. Only on Model 2 (or 2*) has the property been reduced to, or identified with, any microstructural property. On the other models it has not. Moreover there seems no compelling reason to favour Model 2. Scientific practice does not point in favour of it, and even if it did, it would do so only after philosophical work has been done. The fact that some modern scientists, or even most of them, say things such as "we have learned that solidity is XYZ" is not decisive. Such sayings can be taken as elliptical for statements such as "we have learned that the explanatory basis (or causal grounds) for solidity is XYZ". If it turns out that there is no single microstructural basis that is the causal basis for the causal powers, then this eventuality is well handled on model 3. This model requires that there is some basis for the causal powers, not that there is a unique basis. Model 2 on the other hand, requires a unique property. The only possible way to handle the eventuality of multiple realisations, given this account, is to say that there are different kinds of solidity each with its own unique basis. Presumably, each kind of solidity is possessed by one of a limited range of objects. In the case of temperature, it would seem that the physical basis must be different for gases, as against liquids, solids, and sub-atomic processes.

Multiple realisability of states such as temperature and solidity and potentially colour, might be handled by a modification of model 2. It could be that in our reconstruction, we relativise the concept so that temperature is relativised to a range of objects, being that state which, for that range of objects, plays the distinctive causal role for that property. It is plausible to say, for example, that any eye is an eye for an organism of a certain kind. In each type of organism the eye plays a similar causal role, but it is realised by different structures in each type of organism. On this way of thinking, an eye for a spider is one thing, an eye for a human another, but what makes them both eyes is the kind of causal role they play (specified in a formal, abstract way).

The lesson the example of solidity teaches us this. First, we only have reduction or identification if the original concept is of a certain type. If it is not, if it is, for example, a pure dispositional property, then we do not have reduction or identification. In such a case, we might treat the original concept as replaceable by a revised concept. The justification would be, for example, that not only would nothing important be lost by the change, but scientific practice, say, would be enhanced. It would then be possible to have replacement plus identification (or reduction). However, that would require taking the new concept to be of a certain type, e.g., as in Model 2 or 2*, rather than of a mixed type, as in Model 3. We should not assume that in general the first type is superior to the mixed type, and should expect that it sometimes is not.

Second, if it turns out that there are multiple realisations for solubility or temperature or whatever, the only kind of identification is a relativised one. Such a relativised identification is admissible, but it requires some uniting principle. In the case of solidity, the uniting principle seems to be that of having certain causal capacities and powers. What makes the relevant microstructural property count, in the proper context, as solidity, is that it occupies a certain causal role. In the case of colours such a relativised account, one relativised to observers and the way they appear, would seem to be the most appropriate account.