Q: What is the subject-wise percentage of copied work allowed as per the norms of plagiarism?
Firstly, ‘copied work’ may not be the best articulation of what you are seeking to know. ‘Similarity’ would be closer to the mark. :)
Anyway, to answer your question, there isn’t any industry standard or norm as such for different subjects or subject areas. So too with kinds of articles, be they original research-based articles or secondary research-based. There are norms, or permissible amounts, for the various sections of a paper. The Methods and Materials section, for instance, is challenging to write in ‘your own words,’ as that is the most scientific part of the paper, so some similarity is understandable here. On the other hand, sections such as Introduction and Discussion/Conclusion need to be written from your perspective, and so, the similarity tolerance/threshold for these sections is low. The Results section, of course, would depend on the results of your experiments/research (unless, say, you are doing a reproducibility study).
Overall, you should aim to keep to the similarity or plagiarism percentage around/below 15%. Incidentally, something that can help you with this is the manuscript assessment check powered by R Pubsure (a family brand), which provides a plagiarism check in its Pro version, as you can see below. Through its various checks, R Pubsure aims to help you make your manuscript submission-ready in every way, thus helping you minimize or even avoid desk rejection. You may learn more about R Pubsure here and view a sample check report here.
In fact, if interested in R Pubsure, you may also wish to sign up (free) for Pubathon, a two-day research publishing event presented by R Pubsure with the aim to help you ‘speed through manuscript submission.’ You may learn more here: Pubathon – Speed Through Manuscript Submission
Anyway, hope all that helps. And if you are working on or are about to submit a paper, all the best for reducing the similarity percentage and for submitting it!