In 1869 Berthelot had already reported to have produced oil from coal, but only around 1910, in Germany (for cogent reasons), the subject was taken up in earnest. First Bergius, and then Pier (of the BASF, not fortuitously either) developed what was eventually called the Bergius-Pier process, a two-stage process for the liquefaction of coal. It was Matthias Pier who developed the catalysts for the second stage in which the primary products of the first coal-hydrogenation step needed to be upgraded. When he was done, by the late 1940s, many of the hydrotreating (HT), and some of the hydrocracking catalysts we are familiar with had seen the light of day: Co/Ni/Mo/W (mixed) sulphides, supported on alumina, Al2O3:F, or amorphous silica-alumina. The alumina-based catalysts were obviously ready-made for the hydrotreating of petroleum fractions as well, and therefore they were the catalysts of choice when this family of processes took off after WW II, and they also underwent, of course, further development. But it was only in the 1960s that people really started to imagine how these HT catalysts were structured and how they might function, giving rise to a variety of catalyst models under the impact of new spectroscopic techniques, and eventually leading to the Topsoe model and all that. Now, the question before us is: did the greatly increased knowledge of the catalysts affect in any substantial way their further development? Or is it rather the case that fundamental knowledge only started to be acquired when HT catalysis was already almost mature? This paper, then, has two parts: first, the essential features of catalyst development in the pre-Molecular-Aspects era will be reviewed, and then some salient aspects of the modern (so-to-speak) era will be discussed with a view to assess whether or not catalyst development can be seen to be changing, as it was often claimed, from an art to a science in the latter era. (C) 2016 Elsevier B.V. All rights reserved.