[HIST 2297] Is Pegasus observable?

In Utilitarianism, Mill writes as follows:

[T]he sole evidence it is possible to produce that anything is desirable, is that people do actually desire it.

No reason can be given why the general happiness is desirable, except that each person, so far as he believes it to be attainable, desires his own happiness.

Mill has been accused of making an argument that commits the fallacy of equivocation because the term 'desirable' is ambiguous. More explicitly, it seems that Mill argues as follows: 

1. Whatever is desired is desirable. 
2. Happiness is desired. 
3. Therefore, happiness is desirable. 

Some have argued that 'desirable' is used in a descriptive sense in the first premise to refer to what is able to be desired, and in a normative sense in the conclusion to refer to what ought to be desired.

Now, instead of 'desirable', consider 'observable'. Constructive empiricists, such as van Fraassen, argue that "science aims to give us theories that are empirically adequate." A theory is empirically adequate, according to van Fraassen, just in case "what it says about the observable things and events in the world is true."

Critics (e.g., Kitcher) have argued that constructive empiricists have to abandon belief in prehistorical entities, such as dinosaurs, since their existence is inferred from the fossil record. To this, constructive empiricists respond by arguing that the observable/unobservable and the existent/non-existent distinctions are independent. In support of this claim, constructive empiricists (Muller and van Fraassen 2008) say the following:

Before we know whether Pegasus exists or not, we classify it as observable; it is in part because flying horses are observable that we are so sure there aren't any (original emphasis).

What's the argument here? Maybe it's supposed to go something like this:

1. Whatever can be observed is observable.
2. Flying horses can be observed.
3. Therefore, flying horses are observable.

Is this a valid argument? Or does it commit the fallacy of equivocation? Compared to Mill's argument, is it the case that 'observable' in the first premise is used in a descriptive sense (i.e,. what is able to be observed), whereas 'observable' in the conclusion is used in a normative sense (i.e., what ought to be able to be observed)?