The purpose of this work is to homogenise a number of different data sets of daily values of UV-radiation to enable studies of the temporal variation of UV-radiation and atmospheric processes. Efforts are concentrated on the longest series measured using a Robertson-Berger and a Solar Light Model 501 radiometer. The lack of practical and reliable calibration standards have made measurements of UV uncertain. There are also several sources of uncertainty inherited in the designs of the instruments as well as uncertainties in the input data to the models used for corrections and for filling gaps of missing data. The primary goal is to achieve a homogenous data set. The second goal is to mimic a true CIE-weighted irradiation, McKinley and Diffey (1987). The third goal is to be as close as possible to an absolute irradiance scale. There are some factors that will prevent the achievement of these goals. These factors will introduce uncertainty in various fashions and affect the data differently. The applied corrections are assumed to remove systematic differences in a statistical sense. However, as will be discussed below, large differences may still remain in the hourly data. For those not familiar with measurements in the UV spectral range it may seem surprising that UV data recorded by different instruments differ so much. This is partly due to the very large change (several orders of magnitude) in the solar spectrum combined with small differences in the spectral responsivity of the instruments in this range. Apparently small differences in spectral responsivity will produce relatively large variable differences in the output only by changing the solar elevation during the day or during the year. The yearly effect will be seen in the results below.