The following plots show the ratio of the measured F₂ to theory (calculated with CTEQ5D PDFs) for 96/97 data (red dots), 99/00 data (black dots) and the combined data set (blue dots):
(from Matthias' page)
The consistency of the two measurements can be verified from the pull distribution defined as (F₂^96/97-F₂^99/00)/d, where the d is the combined error calculated assuming uncorrelated errors from the two measurements, d=sqrt(d²_96/97+d²_99/00). The pull distribution width is close to 1 as expected, but there is a small relative normalisation difference between the measurements.

The relative normalisation of the two data sets was studied by multiplying the 99/00 F₂ measurement by a normalisation factor and calculating the mean of the pull distribution as this factor was varied.

A normalisation shift of 2.35% was found to produce the best agreement between the measurements. This is consistent with luminosity uncertainties of 2% in 96/97 and 2.25% in 99/00. The luminosity uncertainties should not be considered as totally correlated due changes in the luminosity monitor filters and acceptance and to varying beam conditions.