I would expect that that would be handled - the experiments (records) are each independent of the others (within a project), so each experiment can be repeated (as in use the same set of ingredients as required), with results noted. The repetition would be evident in the report in which a selection is based/ordered by some factor - like an ingredient by quantity - for a project. However repetitions occur for the range of combined ingredients ensuring no one result is a statistical fluke
It would be unusual to only record a "result" for an experiment and not a series of observations / measurements that then need to be analysed to generate a conclusion. Hence the question in post #13. That potentially could be quite complex, depending on business needs
Projects themselves may have an internal hierarchy which could be supported with an appropriate FK to the Parent Project.
If needed, perhaps an IsRepetition flag or a Repetitionof FK to the original (Parent) experiment in the experiment table could be advised
@The_Doc_Man ? The collation of the results from each I think is basically the same suggested report.