Problem with fitting histogram

I need to fit my histogram which contains Trigger Efficiency

First fit is gaus function, second just a line. Which one function will be good to fit empty space? Or its possible to fit it with one own function?
Here is the code.

#include "TH1.h"
#include "TF1.h"
#include "TMath.h"
Double_t fitf(Double_t *x, Double_t *par)
return 1;
void rysunek3()
   TFile *file = new TFile("histograms.root","open");
   TH1F *licz = (TH1F*) file -> Get("h_TRK50E0_TRK70E0");
   TH1F *mian = (TH1F*) file -> Get("trk50E0");
   licz -> Divide(mian);
   licz -> Draw();
   double x1,x2; 
   int nbins=licz->GetNbinsX();
   for (int i = 1 ; i < nbins ; i++)
     if (licz->GetBinContent(i) > 0.2 &&  licz->GetBinContent(i) < 0.22) {x1 = i;}
     if (licz->GetBinContent(i) == 1.0){
       x2 = i;
  Double_t par[13];
 TF1 *func = new TF1("fitf",fitf,x2,x2+40,2);
 TF1 *g1 = new TF1 ("g1","gaus",x1-20,x2-20);  
printf("x1 = %d, x2 = %d",x1,x2);


it really depends on the problem you are trying to solve here, but the erf function is usually good to fit efficiencies. You can for example have a look to this study of D0 which is one of the first google entries: … ackage.pdf