You can figure out that there is no luminiferous aether between the planets with an interferometer. You can observe that the light of distant stars is bent by the sun's gravity during an eclipse. You can check that the planets appear in the sky where your celestial mechanics predict them to be. You can find out what elements the stars are made of with a prism.
The rest of those are not experiments, they are descriptions of natural phenomena we have no power of repeating. Which goes to show that you can have a science that does not have experiments but does have observations.
We clearly have different definitions of "experiment". Having a theory (e.g. light is affected by gravity), making a prediction (that star that should be behind the sun will be visible) and then doing an observation fits my definition of experiment. It's more difficult to reproduce that measuring the speeds of falling apples, but eclipses are not that rare and you can always come up with other predictions from you theory and do the corresponding observations.
I would rather argue that we can in fact do experiments in economics as well. It's just harder to draw conclusions from the observations because there are lots of variables you can't control for easily.
There are lots of experiments you can do.