(This whole argument can be captured by reading just the bold bits. The rest is packing and details you could look up anywhere.)
Science is, as virmaior points out, historically part of Philosophy. It remains such, and always will. Any discipline hopefully becomes a more effective version of itself over time, and philosophy has done this by including science. Science is more effective than older philosophy. But so is the rest of philosophy. No part of something can be uniformly better than the whole of it. There is space around the part that remains to be considered.
We have a tacit understanding that the internals of any given science are best left alone by philosophy, to the degree they are working and not causing damage. And we refer to only 'what is left over' after partitioning off science as 'philosophy per se'. But that is a polite convention, and not a fact.
Every science is more refined than philosophy as a whole, in that it presumes certain truths which are part of what Kuhn calls its paradigm, and below that, it assumes a group of meta-paradigmatic truths that let it better agree with other sciences -- basically, that anything useful can be approximated well enough in a certain brand of physicalism that all science can be tested with other science.
We know that philosophy is not totally covered by the range of existing sciences because as areas of thought establish enough data to base a paradigm, work being done by philosophers becomes the content of new sciences. Psychology was philosophy at the start of the careers of Wundt and James, and a science at the end. It is hard to argue that this discontinuity changed philosophy, surgically sealing off the domain, or that the work of psychologists itself became totally different by establishing its independence.
But a certain set of principles was established enough that further philosophising in that context was temporarily ready to uniformly accept them. Everyone agreed that if you violate those principles, you are probably wrong. So we had, for the first time, something against which we could test this kind of idea fairly directly. And voila, philosophy becomes science.
As the emergence of new sciences points out, a science, is at root, only more testable than other parts of philosophy, because it has a paradigm, and it is only more refined to the degree that it can look closer and see more clearly when it restricts the breadth of its view by accepting a paradigm.
The refinement of being scientific, which is gained by paradigm establishment, is meant to improve the efficiency of science, and I think that we can all agree that that really works. Modern sciences can move forward cleanly and efficiently. But efficient and effective are not the same thing.
Sciences use up the predictive range of their paradigms, or slowly come to admit they cannot work out some internal inconsistencies which make them less and less convincing. When a science's paradigm becomes overtaxed or weak, the science founders, and stops being efficient. At those points a science needs to fall back on the rest of philosophy to reshape its ideas and give it a new paradigm. Historically, this happens with great regularity. Kuhn calls this a 'revolution'.
When we re-injected the notion of atoms into physics to explain Brownian motion and thermodynamics, those 'new' ideas came from very old philosophy. And philosophy had developed, over time, many of the arguments that let physics make peace with this new perspective and integrate it. Atomists had existed all along, just banished from physics (sometimes literally, q.v. Boltzmann)
I would argue that as information accumulation increases, this refinement of the paradigm happens more and more often, at a smaller and smaller scale, and we don't see as much of the disruption that the word 'revolution' implies. Recent 'revolutions' like relativity caused much apprehension and dynamic interest, but preserved almost all of the existing knowledge.
Since the rest of philosophy bears up science when it becomes confused, without the rest of philosophy, it would eventually fail absolutely. The sciences really cannot be declared more effective than the very thing that makes them up, and upon which they rely to save them when they get stuck. And that thing is 'philosophy per se'.