With the exception of some branches of chemistry and physics, many sciences don’t have the luxury of direct observations.
They must rely on the observed consequences of what has occurred, build a theory to support those observations, and then back check the theory with current observations
History, climatology, paleontology, linguistics, and most others all have this feature.
Evolutionary biology is actually in better shape than most since (despite your protests) there are many direct observations that can and have been made both in the lab ad in the field.
See observed incidences of speciation
Additionally biochemistry leaves a record in the DNA that is as clear as layers of rock to a geologist.