Hello, as a historian with little formal training in statistics (read none) I might be asking a silly question, but from what Iíve researched I think I may be able to use a t-test for the case below. I just want to see if anyone catches any mistakes or sees any way to refine it.

My data are a collection of archaeological sites from various time periods located within a 5 km radius of another collection of hillforts. I want to compare the rates of change in the number of these surrounding sites from period to period (Roman to late antique to early medieval) with the rate of change in the percentage of hillforts (from the total number I have collected) that have at least one of these surrounding sites in each period. Both of these rates of change are expressed as percentages. So for example, between the Roman and late antique periods there is a 70.2% drop in the number of surrounding sites, but only a 19.7% drop in the number of hillforts with at least one of these sites within 5 km. The gap between these different rates of change seemed important to me since they are related changes so I looked for ways to quantify this. I came across a two sample t-test on percents as a way to compare these two values since they have different sample sizes (in the example above, the number of surrounding sites from the Roman to late antique periods drops from 477 to 142 and the percentage of hillforts with at least one surrounding site drops from 95.3% to 75.6%).

So my main question is whether a two sample t-test can be used to compare these changes in percents. More importantly, what would I use for the sample size? For the example above, would I use 477 as the sample size for the drop in surrounding sites from Roman to late antique periods, and 86 as the sample size for the drop in percentages of hillforts with at least one surrounding site since this is the total number of hillforts I am deriving this percent change from?

Any help or guidance would be really appreciated, thanks!