Reply to “Mondrian vs Rothko: footprints and evolution in style space”
(Note - images come from the web and are not meant to faithfully represent the canvases they are merely provided as an illustration)
The Software Studies Initiative at UCSD has produced an information visualization comparing the development of Mondrian and Rothko’s canvas painting in terms of brightness and saturation.
Information visualization, Mondrian vs. Rothko by UCSD (see original UCSD post for explanation)
The results are curious and appear to show a particular pictorial development in the two artists work. Visualizations such as this are useful, certainly in terms of wrestling art history out of its cozy setting amongst endless lines of printed text, and the results seem to document clearly certain avenues of inquiry. At their core visualizations are epistemological and that is no bad thing.
However closer inspection of this particular visualization shows some quite serious problems with the methodology and therefore the end hypothesis. For me this visualization is supremely simplistic and actually skews the artistic development of both painters to the point where it incorrectly re-imagines their work.
Central to this is the choice of data set –
We have selected particular periods in the career of each artist which are structurally similar. In the beginning of a period each artist was imitating his predecessors and contemporaries. By the end period each developed his mature style for which he became famous. In between, each gradually moved moved from figurative representation to pure abstraction.
This is patently incorrect for several reasons. Firstly taking the two painters as separate for a moment, in the case of Mondrian, the 13 year period selected does not represent what is claimed. Mondrian had not reached his mature style by 1917 and what is represented by the visualization as being some form of end point is actually the very beginning of his experimentation with abstraction. His mature style would not develop until the 1920’s.
Comparing the work of his mature abstract style with the experiments of 1917 and a few years before in terms of brightness and saturation is revealing. I don’t want to try and second guess image processing software but some differences are obvious. The geometric colour planes in the earlier pictures are muted compared to the vibrancy of the later works, which suggests the pictures of his mature style would to tend group further up the y axis than is implied. Of course they never became anything like as saturated as Rothko’s red canvases but to only include data up to 1917 unfairly exaggerates the difference.
Composition II - 1920 Composition A - 1920 Composition with red -1930
On a more specific note if the researchers had included Mondrian’s work up to 1919 rather than 1917 in their data set they would have necessarily included the checkerboard pictures whose brightness and saturation seems to contrast directly with the hypothesis of the visualization.
Composition checkerboard dark colours - 1919
In addition it is questionable whether the term ‘mature style’ is helpful when assessing Mondrian’s work which was always teleological thus to speak of a ‘mature style’ is somewhat clumsy. In this context any account of his mature style must include the New York pictures of the 1940’s.
The next problem is that the visualization does not, as is stated, compare periods in the artists’ careers that are ‘structurally similar’. It does not compare like with like. Broadly the Mondrian period presented is his early career and the Rothko period his mid-career. The Mondrian period is one where he begins painting landscapes in an expressionist mode and ends on the cusp of abstraction. The Rothko period begins with the cusp of abstraction and ends somewhere in the middle of his abstract expressionist phase.
“In between, each gradually moved moved from figurative representation to pure abstraction”
To me this seems to be staggeringly misleading. The Rothko pictures from 1944 are not exactly what you would call figurative. Whilst perhaps based on objects they are not figurative in the way Mondrian’s early landscapes depict a specific observable scene from reality. It is possible to compare like with like because early in his career Rothko did paint figures and a structurally similar period would have been to start with Rothko’s work between 1925-35. A casual glance at Rothko’s pictures during this period shows them to be darker, comprising mostly dark greens and browns, and actually in terms of the parameters of the visualization the period where both men were painting figuratively seems to be remarkably similar and not totally the opposite as is implied.
Rothko Ritha of Lilith - 1945 Mondrian Red Tree - 1908
A further problem with the data set is an ambiguity regarding what is included and what is excluded. Looking at Rothko’s catalogue raisonne shows that between 1944 and 1957 he made approximately 370 pictures. Of these 370 only 151 are included in the visualization. On what basis were the 151 chosen and in what sense are they representative of artist’s development during this period? The general impression from the catalogue raisonne is that Rothko did not start consistently painting darker images of the kind represented by the large red circles until the 1960’s.
In light of all this some of the analysis given by Manovich and Carey seems misjudged.
The visualizations also show how Mark Rotho - the abstract artist of the generation which followed Mondrian’s - was exploring the parts of brightness/hue space which Mondrian did not reach (highly saturated and bright paintings in the upper right corner, and desaturated dark paintings in the left part).
Because like for like are not compared this seems plausible but if the visualization presented data from periods that were similar it’s unlikely such a point could be made. It is surely impossible to conclude that Mondrian did not reach certain parts of the brightness/hue space if 30+ years of data is excluded? Given a prior knowledge of Mondrian’s work this conclusion could be speculated but it is not proved by the visualization. In any case looking at Mondrian’s New York pictures are we really expected to believe that he never made bright saturated canvases?
New York City 1 - 1942 Victory Boogie Woogie - 1944
Manovich and Carey continue:
This visualization reveals another interesting pattern. Rothko starts his explorations in late 1930-1940s in the same same part of brightness/saturation space where Mondrian arrives by 1917 - high brightness/low saturation area (the right bottom corner of the plot). But as he develops, he is able to move beyond the areas already “marked” by his European predecessors such as Mondrian.
The immediate problem here is that the visualization for Rothko’s paintings begins in 1944 so any kind of analysis must begin from this date. To speak of his explorations during the late 30’s is not possible with the data presented. Beyond this blip however is a more fundamental problem. To say Mondrian had arrived somewhere in 1917 is borderline absurd. He had arrived at the start of his foray into abstraction but to arrive at the start is not to arrive in the sense meant here where it implies an endpoint. Mondrian’s endpoint would not occur for another 30 years.
In fact the visualization, once mentally adjusted to compare like with like, shows quite clearly that both men began their path to abstraction with bright de-saturated pictures. With prior knowledge we might speculate that Mondrian for the most part, but not exclusively, stuck with this formula whilst Rothko went on to experiment heavily with colour. This however is not a conclusion that can be directly drawn from the visualization.
The notion of pictorial space ‘marked’ by previous generations is interesting and visualizations such as this provide some kind of torch with which to navigate.
But again I wonder if there is a problem here with not comparing like with like. Rothko seems like an obvious candidate for investigations into brightness and saturation but Mondrian less so due to the self imposed reductionism of his palette. Might it not be more appropriate to compare Kandinsky’s work with Rothko’s in terms of colour saturation? And if we did could we possibly say that ‘as he develops, he [Rothko] is able to move beyond the areas already “marked” by his European predecessors such as Kandinsky.’? It doesn’t seem quite right to conclude Rothko moved ‘beyond’ the area marked by Mondrian without taking into account both artists approach to the grid.
The visualization, however, does seem to be useful in certain respects and the patterns do provide a few insights. I mentioned earlier the teleological nature of Mondrian’s painting. This is usually conceived in terms of formal relationships between pictorial elements rather than colour space but the visualization does seem to indicate a parallel between the two. By contrast the pattern suggests Rothko’s work does not proceed with the same purpose. This conclusion must remain totally provisional however due to the lack of Mondrian data after 1917.
There is then a whiff in all this of selecting data that produces a discernible pattern regardless of whether that data is appropriately comparable. Questions remain over what data was selected and why and I think if researchers are going to present such work they must also supply what underlies it. One of the essential features of data is that it can be manipulated and in a discipline traditionally resistant to computational analysis transparency is key.
Information visualization tends to present results as definitive when in fact it is simply just another argument - it needs be open to criticism.
