There are valid quibbles and exceptions on both counts. Some breeds of cats make vocalizations that can reasonably be described as "barking", and water will burn if there are sufficient concentrations of either an oxidizer much stronger than oxygen (such as chlorine triflouride) or a reducing agent much stronger than hydrogen (such as elemental sodium).

In the general case, though, water will not burn under normal circumstances, and most cats are physiologically incapable of barking.

The point of the quote is that objects and systems do have innate qualities that shape and limit their behaviour, and that this effect is present in social systems studied by economists as well as in physical systems studied by chemists and biologists. In the original context (which I elided because politics is the mind killer, and because any particular application of the principle is subject to empirical debate as to its validity), Friedman was following up on an article about how political economy considerations incline regulatory agencies towards socially suboptimal decisions, addressing responses that assumed that the political economy pressures could easily be designed away by revising the agencies' structures.

Relevant.

pfsch. You can burn water if you add salt and radio waves. Or if you put it in an atmosphere containing a reactive fluorine compound. Etc etc etc.