You skim those papers to verify that the data is indeed there and then you cite them as a source of the data. If you have the time or they are otherwise relevant, read the papers to figure out that the data is suitable for your purposes and there are no glaring problems there.
Sometimes, everyone cites something because everyone else has been doing so, but the citation might be wrong (in detail or completely).
An explicit citation is useful, because it lets you and others verify the result.
Checking that the cited paper includes what you suppose it includes is even more helpful, since it might just be a rumour that the thing is included there.
Reading more of the paper would be even more useful, since it might have qualifications that have been omitted by the citing literature, or methodological issues, or whatever else. Mentioning these with the citation would be good.
Some papers that cite the original might contain errata, important qualifications, etc. Hence, it might be worthwhile to check what cites the classical papers, especially if there is something by the same author(s) on the same idea.
I think all of the steps above increase the value of the work, but all of them also take time. Up to you to decide where to cut off the process and how far to go in vetting the data-to-be-used.
But consider also the next new (PhD) student who would like to use the same data? What would you like to have known or had when using it? If it was non-trivial to track down the data, then certainly provide a reference, for example, so that the next generation will have an easier time.