Summary. When describing how much greater air pollution in London was in previous centuries, Lomborg presents a graph pieced together from sections of different datasets. His choice between different datasets covering the same period bears no relation to their likely robustness, and seems to have been deliberately made to increase the impression of a steep decline during the 20th century. Indirectly calculated figures from before 1900 are without justifiable reason multiplied by an arbitrary factor of about 5 in order to fit values around 1900 to the artificially increased steepness of the curve during the 20th century. This deliberate manipulation makes it look as if air pollution was much worse in former centuries.

     When Lomborg wants to present an optimistic view on air pollution in cities, he can do that by claiming that the problems were much worse in the past. To demonstrate this, he uses London as an example (he could not have used a city like Paris, because Paris was heated by wood, not by coal). On p. 164 he writes: "Air pollution is not a new phenomenon that has got worse and worse - it is an old phenomenon, that has been getting better and better, leaving London cleaner than it has been since medieval times."
     His basis for saying so is a paper by the British environmental scientist Peter Brimlecombe. Brimblecombe used an existing computer model which calculates air pollutant concentrations on the basis of the amount of pollutants emitted, the size of the city, and wind speeds. When appllied to cities of varying sizes at the present time it gives values which lie within 20 % of the mean values measured at air pollution monitoring stations in those cities. So it is fair to assume that it could tell us former pollution levels with nearly the same precision.
    Brimblecombe recorded the size of London in various years, the amount of coal burned for fuel within the city limits (this is known), and the sulphur content of the coals used (estimated from geological knowledge of the coalfields). With this input to the model, he got the results which are presented in the figure shown below.

We see that the pollution, e.g. with sulphur dioxide, peaks near the end of the 19th century, which agrees with data showing that the number of  days in London with smog increased  much during the latter half of the 19th century. We also see a steep drop in pollution levels from about the year 1900. The model produces this drop because the diameter of London grew much at this point of time, which would have the effect of diluting the pollution. But this does probably not fit with what actually happened. Brimblecombe writes:  "The decline shown in the pollutant concentrations  . . seems particularly large, and is probably exaggerated by the diffuse nature of the outer parts of the city at the turn of the century. " But unfortunately, to support his point, Lomborg emphasizes this decline and does not cite the author´s reservations.
    Next, Lomborg continues the curves up to modern times by fitting in data of actual measurements of the air in central London from about 1930 onwards. Concerning sulphur dioxide, these data are taken from a paper by D.P.H. Laxen and M. A. Thompson, published in 1987. Below I present a figure from their paper with those curves that were used by Lomborg.
 

We need to go into some detail with this figure. First we notice that in the beginning the measurements are neither systematic nor continuous. And next we notice that sulphur dioxide has been measured by two different methods, called A and B. In general, it seems that method B gives higher values than method A.
      Now when we compare this figure from Laxen and Thompson with the SO2 curve in Lomborg´s figure 86, we see which parts Lomborg has used. Starting from recent time and moving backwards, we recognize the course of Laxen´s curve 9, then curve 6 (the lowest of four parallel curves), then the two separated pieces of curve 4, and finally, in the oldest part of the diagram, curve 2, i. e. the upper of three curves. So to the right in the figure, he picks the lowest curve possible, and to the left in the figure, he picks the highest curve possible. He does this even though curve 2 is based on fewer measurements than curve 1; the prominent high peak at about 1935 is actually based on quite few measurements. Furthermore, the choice of curves is inconsistent. To the right he uses curves based on method A, but to the left curves based on method B, which, as noted above tends to give higher values. By this kind of selectivity - which is unacceptable in scientific work - he manages to make the slope from the 1930s till now steeper than it would otherwise be. Again: it seems that Lomborg has an intention to make the decline in pollution look as large as possible.
     The next trick is the way that the curves from the two different studies are combined. Brimblecombe´s curve ends in 1929 with values slightly below 100 µg, whereas Laxen´s curve, in Lomborg´s version, starts around 1934 with a value about 400 µg, i.e. 5 times as much. So Lomborg assumes, without any actual knowledge about this, that levels in the city center are 5 times as high as the average values over the whole city, because with this assumption he can make a smooth junction of the two curves. In note 1163 Lomborg writes how he has done this: "This necessitates an upwards adjustment of Brimblecombe´s data by approximately a factor 4." (well, 5 is approximately 4, isn´t it?). So whereas the maximum concentration per m³ of SO2 in the late 19th century is given by Brimblecombe as around 180 µg, this is increased by Lomborg to 900 µg, as we see in his figure. So actually he has multiplied the whole curve by a factor 5. He defends this manipulation as follows in note 1163: "However, since Brimblecombe´s model data are averages for London as a whole, it must be assumed that central London was much more polluted, in line with actual measurements." But this reasoning is flawed. London in earlier times had a smaller diameter, with shorter distance to unpolluted surroundings, so the difference between the air quality of the center and the average for the whole city has not been constant. Brimblecombe also gives data for Epping Forest, about 20 km from the centre of London. Here we had 70-125 µg of SO2 per m³ around 1970, at a time when the values in the center of London were 200-250 µg per m³. So it is not true that the values in the center are up to 5 times higher than the average for greater London. The maximum centre/average ratio that one could justify is a factor of somewhat less than 2.
    Remember that the right part of Brimblecombe´s curve probably declines too steeply, and that the values after the year 1900 should probably have been higher. Taking this into consideration, it would not be necessary to multiply the curve by as much as 5 in order to make it fit with the actual measurements. And remember also that the left part of the curve from Laxen has been made steeper by Lomborg. Only by multiplying the Brimblecombe curve could he make its slope fit in with the artificially increased steepness of the curve from 1931 onwards.
     In any case, Lomborg purports to use only "the facts" - the actual figures - so by his own standard he is not allowed to multiply the figures by 5 when it happens to suit him.
     If the curve had not been multiplied by 5, what would Lomborg´s figure have looked like ? In that case we would have had a concentration of about 180 µg around 1850, and a concentration of  300-400 µg around 1935. So the curve would then have shown that pollution was less in former times, and had increased until shortly before the big smog catastrophe in 1952. In that case, Lomborg would not have been able to back up his assertion.
    Lomborg´s statement that ". . the levels of the 1980s-1990s are below the levels of the late sixteeenth century" are definitely false, because even if we assume that a multiplication of Brimblecombe´s curve might be justified for those time periods when London  had a large diameter, this would certainly not be justified for the early years, when London had a small diameter. That is, the values given in Lomborg´s figure for the time before and around 1600 are certainly very wrong.
     In short, Lomborg´s postulate -  that air pollution was much worse from 1700 to 1900 than after 1900 - is most likely not true, and his postulate that London in the 1980s-1990s was cleaner than in medieval times is definitely unjustified.  To multiply figures from one period by any factor that one likes, in order to make them larger than those of another period, is gross manipulation.
    Some people think that even though Lomborg´s curves are not correct, this does not change the big picture which is that the air quality of London today has improved greatly relative to what it was. In their view, this is heartening to know, and Lomborg is right in telling this good story. However, they fail to see that Lomborg has an agenda: He wants to give us the impression that the air quality has improved gradually from about 1850 "by itself", and that this improvement bears no relation to the passing of the Clean Air Act in 1956 (p. 170). However, had his curves not been manipulated, we would have seen that air pollution culminated in 1935 and 1952, and that the situation improved only when the Clean Air Act was passed. So the agenda is: does it, or does it not, help to combat pollution by inforcing restrictions? If we accept Lomborg´s view - that it does not help - then citizens of many large cities around the world will not be able to experience the same heartening facts that the citizens of London experience today.