**5. Other Universals**

BFO divides continuants into independent, specifically dependent and generically dependent continuants. Independent continuants are either material or immaterial entities, of which the former include common objects like tables, elephants and the like. The latter divides into *continuant fiat boundaries* and *sites*, which support the ontology of things like holes and geographic boundaries [8] (p. 41). Following the process discussed above, all these can be matched straightforwardly from the trope ontology to their respective universals in BFO. For example,

*HMAS-Beagle is part of Sydney-harbour.*

*The hold of HMAS-Beagle is part of HMAS-Beagle and a hold.*

*A hold is a continuant-fiat-boundary.*

The example would be represented by the following predicates, readily seen to be following the same matching principles for universals described above.

**p**([1], part, 'HMAS-Beagle', 'Sydney-harbour').

**p**([3], part, hold('HMAS-Beagle'), 'HMAS-Beagle').

**p**([4], part, hold('HMAS-Beagle'), holds).

**p**([2], part, universal(hold), universal(hold)).

**p**([5], part, holds, 'continuant-fiat-boundaries').

**p**([7], part, universal('continuant-fiat-boundary'), universal('continuant-fiat-boundary').

One aspect about such matchings does need clarification. Namely, the trope ontology would need extra predicates to capture the rules that are part of BFO. For example, material entities cannot be part of immaterial entities. Since the trope ontology is primarily expressed as a logic program (i.e., Prolog), it is a relatively simple matter to add extra predicates that capture such rules. One approach is to implement the rules as constraints that can be checked for an entity. For example, below is a possible constraint on any entity that if it is part of *immaterials*, then it cannot also be a part of *materials*.

*constraint(X)* ←− *part(X, immaterial), not( part(X, material))*.

*Specifically, dependent continuants* are either a *quality*, such as "the mass of this piece of gold" or "John's being the biological son of Jane", or a *realizable entity*, which are either a *role* or a *disposition*. Realizable entities are like qualities but are not always part of the entity that they adhere to. For example, "John's ability to sleep" is a disposition, because John does sleep, but not always (assuming John is an ordinary person). One can see the same with roles, which are dependent continuants that an entity has due to circumstances and which leave the bearing entity physically unchanged if the role is removed [8] (p. 58). For example, John's being CEO would be a role. *Generically dependent continuants* are like their specific cousins but can be copied between bearers. For example, the arrangemen<sup>t</sup> of chess pieces on a board can be exactly copied to another chess board.

The implementation of such dependent continuants is structurally simple through the use of functional terms. Previous examples, such as "the hold of HMAS-Beagle" already show the use of functional terms. Dependent continuants such as qualities would use functional terms as well. For example, "the temperature of 37-C is part of john" in predicate form would be p([1], part, temperature('37-C'), john). However, qualities such as these are only part of their bearers in the most general interpretation of parthood. That is, we can only say in the most general sense that John's temperature is also part of the mereological sum of all humans (of which John is part). We will probably want more nuance to reflect the nature of dependency. The trope ontology already has an example of generic dependence in the case of informational states. There, we use reserved terms to indicate the special nature of the entity. That in turn enables particular treatment (i.e., logical rules) of informational states. On that foundation of informational states, the trope ontology has previously defined roles as entities in some description (i.e., information) that classify entities in the world [11] (p. 182). However, such an approach is more nominal than in BFO, because it relies on the entities with the information for the role to exist. Nonetheless, the same approach of special terms to represent qualities can be used. For example, we might preface qualities with "quality of", as in "the quality of the temperature of 37 ◦C". This would enable us to recognize qualities for what they are and assure correct inferential rules. This approach follows Eschenbach and Heydrich's use of restricted domains, mentioned earlier.
