Or they didn't. If it's real, why not prove if it's real?
If it was part of an experiment published 20 years ago, and it was conclusive that dolphins can think, presumably we would have heard of it by now.
And btw, while I'll accept the point in theory, do you have any realistic examples of unrepeatable experiments in mind? Besides ones involving people, I guess.
So when that animal dies the experiment cannot be repeated and the research is pretty much over until they find another smart animal to work with.
So many of these animal experiments are very non-repeatable both in theory and in practice.
Another anecdote to the mix:
A researcher wanted to do extensive, repeatable tests with dolphins to test their abilities in communication using a symbolic language. He'd gone to a lot of work designing a board that dolphins could "read" using echo-location, and designing a language (in the technical sense) with syntax, grammar and semantics.
When he described and demonstrated his experiment to the dolphin keepers, they were encouraging and supportive, but said it wouldn't work. When asked why, they said the dolphins would get bored and stop participating.
The researcher tried anyway, and initially got a great deal of success. Early results were promising, but needed repeating to make sure there were no statistical flukes (if you'll pardon the pun) or coincidences.
When he tried to repeat the experiments the dolphins refused to cooperate. When he tried harder to get them to cooperate, they physically swam under him, liften up, and pushed him out of the pool.
The dolphin keepers had successfully predicted the dolphins' behavior. They had a theory, which let them make predictions, which the researcher inadvertently tested and showed to be accurate.
Did it get written up? No. The researcher thought his experiments were a failure because he couldn't get the repeatability. The dolphin keepers weren't scientists, they didn't write anything up.
All this was reported by the keepers in a TV show I saw over 10 years ago, but it suggests several things.
People who keep dolphins believe them to be intelligent.
That belief leads to correct behavioral predictions.
Some researchers have trouble doing experiments, because the dolphins appear to refuse to cooperate on boring tasks.
Additionally, given that dolphins can apparently teach each other behaviors the entire batch of dolphins they've got in their research tank is contaminated by the first dolphin.
Designing experiments to determine if your test subjects are smart enough to mess with you is hard and it's harder if they all live together and gossip between sessions.