I am absolutely sure that the following which could not be achieved by using science models is proving that an object exists. Unfortunately, you can often come across a lot of information that consists of things which can't be explained by they are in use. It can be either abstract points or objects which state of aggregation can't be measured.
And what about the second one, the scientist's next step, if the experiment shows that the scientist’s hypothesis is false should be :Make a new hypothesis and run a new experiment. When something goes wrong, scientist use the same data but in other way, trying to improve the obtained datum, but if it doesn't fit they don't give up and go on making new experiments :)