The correction, validation and homogenisation of the long temperature series of Padova (1725-1998) are discussed. After a careful historical investigation made in order to discover the metadata concerning the series, i.e., instrument features, calibrating methodologies, operational procedures (i.e., exposure, timing and number of daily observations), instrument maintenance, relocations and instrument replacements, the series has been corrected for all the systematic errors derived from any change in the instruments or the operative methodology. Above all, correction focused on instrumental drift, scale expansion, building influence, relationship between indoor-outdoor measurements, minima and maxima evaluation from observations performed at different times, homogenisation for difference of level and change of site. Statistical tests applied to the data and the comparison with other known series has clarified some uncertainties about exposure and operational procedures that the historical analysis of metadata was unable to solve. Moreover, gaps have been filled after the comparison of the series with others of neighbouring sites. The critical work of debugging, correcting, validating and homogenising the series is essential for a correct interpretation of data, as in some cases the errors that have been corrected have been found to be greater than the climate signal. Especially in the early period, the algebraic sum of the corrections of the mean daily temperature exceeds 8 degrees C, where monthly corrections can reach 6 degrees C. After correction, validation and homogenisation, the linear trend of the Padova series is positive, +0.31 degrees C/100 yr over the period 1774-today. Looking at post-industrial warming, the temperature rise is +0.44 degrees C in the last 130 years, which means +0.34 degrees C/100 yr, not far from the above bulk average.
