I fit run4 data with Maxwell func., then integrated the function within the bin range.

The ststistic error of the integral was obtained from IntegralError() function of TF1.

The systematic error is obtined from the paper.

    Double_t run4dNdydpt2pipt = fmaxwelldNdydpt2pipt->Integral(Low, Hi);
    Double_t run4dNdydpt2piptStatErr = fmaxwelldNdydpt2pipt->IntegralError(Low, Hi);
    Double_t run4dNdydpt2piptSysErr = run4dNdydpt2pipt*0.094;
    Double_t run4dNdydpt2piptTotErr = TMath::Sqrt(run4dNdydpt2piptStatErr*run4dNdydpt2piptStatErr+run4dNdydpt2piptSysErr*run4dNdydpt2piptSysErr);
 

The integral gives 0.0236 and total error is 0.0022.

For my analysis, the value is 0.0156 and total error is 0.0048.

So the sigma (difference /sqrtr(run4 error^2+run10 error^2)) = 1.18

Actual output :

Low : 1.7 Hi : 2.5 <--pt range
integral: 0.02355 Stat. Error: 0.000313937 stat + sys: 0.00223585  <--for run4
Run10integral: 0.0155785 Stat. Error: 0.00308793 stat + sys: 0.00476459  <--for run10

Difference: 0.00797156 combine error: 0.00673815 sigma : 1.18305  <--their difference

The soilid line at the last bin shows integlated value for run4 and dashed line shows sys+stat error.