I think this is a more complex task, as ‘resemblance to a (harmonic, or whatever) series’ would depend on a multidimensional measure. as Antonio says, you need to consider cardinality, but also you need to determine what harmonicity is based in, and as Karim points out, what’s the ‘aproximation’. but on top of that you may need to decide something that implies psychoacoustics as well. how do the following three series compare?
A. 1fn, 2.1fn, 2.93fn, 4.1fn, 5.09fn
B. 5fn, 7fn, 9fn, 11fn
C. 1.111fn, 2fn, 3fn, 4fn, 5fn
as can be seen, serie A is inharmonic, B is harmonic, and C a somewhat special case. how would you measure its (dis)similarities? from the psychoacoustical point of view you can address Terhardt and Parncutt, and you if you want to extend this to pitch class sets just go for Morris or Lewin.
sorry for not giving an accurate citation, but a web search will suffice.
pm me if you need more details.