Your browser (Internet Explorer 7 or lower) is out of date. It has known security flaws and may not display all features of this and other websites. Learn how to update your browser.


An End of All Explanations: Visualizing and Interpreting Lewis’s Revisions to Part IV, Chapter 9 of Tarr

In my last post, I described a working hypothesis about Wyndham Lewis’s revisions to Tarr: that in producing the 1928 version, he was principally interested in adding voices to the earlier 1918 version, particularly through the addition of FID.

I decided to test this hypothesis by looking at one of the more disturbing chapters in Tarr. Chapter 9 (Chapter 8 in the 1918) of Part IV begins in Kreisler’s apartment following the rape of Bertha. It begins by narrating the aftermath of the rape, and as the chapter progresses it narrates retrospectively the rape itself. Already in the 1918 edition, this chapter is told primarily from Bertha’s perspective. But I was interested to see if the 1928 gave more of her perspective—particularly to see whether Lewis increased his use of FID to render her thoughts.

I proceeded in the following way. First, I used Juxta Commons to collate plain text versions of the chapter in each witness. I then used the Juxta Commons TEI export feature, which outputs a TEI file with all variants identified in <app> and <rdg> elements. I then marked up all variants that I considered significant to the question of voice. For each such variant, I added three descriptions: whether the 1928 revision added or removed an instance of character speech; whether it added or removed direct, indirect, or free indirect discourse; and which character’s speech had been removed or added. I also recorded instances in which narration about particular characters was removed or added. I used the same simple TEI markup scheme to record all of these features: I added a “type” attribute to every rdg element corresponding to a significant 1928 variant, and specified three attribute values: either “add” or “remove”; then either “direct,” “indirect,” “FID,” or “narration”; then either “Bertha,” “Kreisler,” or “other.” Once the TEI file was thus marked up, I opened it in the Mandala browser for visualization.

There are a few drawbacks to this approach. First, just because I recorded my interpretations in TEI, and just because they generated graphs, the findings are by no means “objective,” or “scientific.” The results show my own personal, subjective reading of the variations between the two witnesses—and of course I went in to this process looking for changes in voice, and already with a working hypothesis in mind. In order for these readings to become more properly objective, a much greater number of annotators would have to be used. Second, my markup scheme is quite crude. For example, the “add” attribute value doesn’t distinguish between character speech that is entirely new to the 1928 version (i.e., completely new dialogue that didn’t appear in the 1918) and character speech that is merely changed from one form of discourse to another (for instance, a passage that was indirect in the 1918 and turned into FID in the 1928). Further, in the latter case, my markup scheme doesn’t say which form of discourse was replaced in the 1928 revisions. My “remove” is thus not strictly the opposite of “add,” since it only specifies character speech that was completely removed from the 1928 version—not character speech that was changed from one form of discourse to another.

These warnings aside, I believe my results are still significant—and that they support my provisional hypothesis.

Consider the following Mandala visualization:

This graph shows every instance of character speech that I identified as being added or removed by Lewis as he made his 1928 revisions, and shows whether these additions or deletions were carried out on Bertha’s or Kreisler’s speech. It shows that the vast majority of such activity concerned Bertha: Lewis removed 11 instances of her speech and added 55; he removed none of Kreisler’s and added ten.

The results become more interesting still when looking at the particular types of character speech Lewis added and removed. Consider the following, which shows all additions of FID in the 1928, and specifies whose FID Lewis adds:



This shows that of the 41 instances I noted in which Lewis adds FID, 39 of them add Bertha’s FID, and only 2 add Kreisler’s.

Here is one example of a passage in which I indicated added Bertha FID. In this scene, Bertha is narrating an event that came before the rape, when Kreisler is painting her and oddly remarks, “Your arms are like bananas.” The first passage is from the 1918, and the second from the 1928. Here my highlighting indicates particular variations between the witnesses.

“Your arms are like bananas!” A shiver of warning had penetrated her at this. But still he was an artist: it was natural—even inevitable—that he should compare her arms to bananas.


‘Your arms are like bananas!’


A shiver of anxiety had penetrated her at the word ‘bananas’: anybody who could regard her arms in that light was inartistic: she was distinctly glad that her ‘good’ legs had not been wanted. He was a modern artist of course and it was natural, perhaps inevitable, that he should compare her arms to bananas.

I tagged “Oh!” as an addition of Bertha’s direct discourse, since it didn’t appear at all in the 1918. I did not tag the “warning/anxiety” variant since it doesn’t clearly present a different type of character speech—although “anxiety” does perhaps bring us further within Bertha’s subjective perspective than “warning,” which may appear more objective to Bertha, possibly indicating detached narration. The longer passage highlighted in blue, however, represents a clear addition of FID. Here the voice of the narrator mixes with that of Bertha: we are clearly taken into Bertha’s thoughts as they unfold (the judgment “inartistic” belongs to Bertha’s, and the staccato rhythm of the passage, indicated by the colon, gives us Bertha’s process of thought as is develops), though the use of the past tense and the third-person pronouns continue to indicate the presence of the narrator. In the concluding sentence of this passage, both the 1918 and the 1928 use FID; since they differ only in details, I didn’t tag the variants. This last sentence does, however, provide a useful reminder that FID—particularly Bertha’s FID—was already a significant part of the 1918 Tarr, and that Lewis’s revisions in many cases simply intensified the effect.

Here is another example of added FID. In this passage, Bertha narrates the rape itself, referring to Kreisler as “it.” Again, the 1918 is shown first, followed by the revised 1928:

It had quietly, indifferently, talked: it had drawn: it had suddenly flung itself upon her and taken her, and now it was standing idly there. It could do all these things. It appeared to her in a series of precipitate states. It resembled in this a switchback, rising slowly, in a steady insouciant way to the top of an incline, and then plunging suddenly down the other.


The figure talked a little to fill in an interval; it had drawn: it had suddenly flung itself upon her and done something disgusting: and now it was standing idly by the window, becalmed, and completely cut off from its raging self of the recent occurrence. It could do all these things: it appeared to be in a series of precipitate states: in this it resembled a switchback, rising slowly, in a steady innocent way, to the top of an incline, and then plunging suddenly down the other side with a catastrophic rush.

The variant highlighted in red, “taken her” / “done something disgusting” clearly moves us further into Bertha’s own subjective perspective: while “taken her” remains somewhat detached, more likely the words of the narrator than of Bertha herself, “done something disgusting” gives Bertha’s own shocked reaction as she struggles to narrate the event. Since both passages were already in FID, however, I did not tag the differences. The passage in green gives new FID recording Bertha’s impression of Kreisler, and I tagged it as such. I tagged the remainder of the passage as added FID, primarily because of the grammatical shift from full stops to colons: as in the previous passage, the colons intensify the effect of watching Bertha’s thoughts unfold before our eyes. While in the 1918 it would be possible to read these analogies as deriving from the narrator, the use of FID in the 1928 clearly suggests that they derive from Bertha herself.

Another significant visualization concerns the removal of narration. The following shows every instance in which narration describing a particular character is removed in Lewis’s 1928 revisions, and whether these narrator’s descriptions concern Bertha or Kreisler.

The chart shows that of the eight instances I recorded as removed narration, all eight removed narration describing Bertha; no narration describing Kreisler was removed. The significance of this can be seen in the following passage, in which Bertha attempts to come to terms with the significance of the rape:

 She now had to go away as though nothing had happened. It was nothing. After all what did it matter what became of her now? Her body was of little importance—ghosts of romantic consolations here! What was the good (seeing what she knew and everything) of storming against this man. She saw herself coming there that afternoon, talking with amiable affectation of interest in his work, in him (in him!), sitting for him; a long, uninterrupted stream of amiability, talk, suddenly the wild few minutes, then the present ridiculous hush.

The moral, heavily, too heavily, driven in by her no doubt German fate, found its mark in her mind. What Tarr laughed at her for—that silly and vulgar mush, was the cause of all this. Well!


Now she had to take her departure as though nothing had happened. It was nothing actually, nothing in fact had happened: what did it matter what became of her? The body was of little importance: what was the good (seeing what she knew and everything) of storming against this person?

Rather than primarily adding, the revisions here mostly remove. A few details do give us more of Bertha’s thoughts: for instance, “Her body” becomes “The body,” the narrator’s third person pronoun replaced with Bertha’s own definite article; and Bertha’s description of Kreisler as a “person” rather than a “man” is given, the latter being a nomination more likely to come from the narrator. But most significant here is the extensive removal of the narrator’s judgments of Bertha. The character Tarr—widely seen as a mouthpiece for Lewis himself—repeatedly complains of Bertha’s penchant for interpreting her life through the lens of German romanticism. In the 1918 edition, the narrator twice steps away from Bertha’s own thoughts to deliver authoritative judgments upon them, linking Bertha’s way of thinking, like Tarr does, to the distortions of a German romantic lens. Indeed, in the final part of this passage in the 1918, Bertha is shown to accept Tarr’s judgment of her. In the 1928 version, these authorial judgments are completely removed, as is the notion of Bertha’s agreeing with them, and we are left only with Bertha’s FID.

Working on the versions of this particular chapter has raised for me a few ethical problems. First is the ethical problem of a male writer (Lewis) trying to imagine a female character’s reaction to a rape. It seems to me that FID presents possibly the “least unethical” narrative technique in such a case. Direct discourse would risk appropriating a voice–mistakenly imagining that Bertha’s perspective could be reproduced fully and accurately. Authorial judgments like those just described seek in a detached manner to describe what cannot be understood from a detached perspective. FID represents a middle ground, giving some of Bertha’s words, but mixing them with the words of the narrator, and so not claiming to inhabit her perspective absolutely.

The other ethical dilemma is mine rather than Lewis’s: the problem of writing in a detached manner about a male writer’s representation of a female character’s reaction to rape, and using technical language and graphs to do so. It is an uncomfortable critical position, without doubt—the kind of uncomfortable ethical position in which Lewis is always placing his readers. But as I believe my investigation is showing, Lewis’s revisions aimed precisely at making his readers more uncomfortable, and less able to resolve ethical dilemmas. By providing more of that uncertain mixture of voices, FID, and by removing more of the authorial judgments by which readers might hope to find some ethical anchor in the text, Lewis made reading the 1928 Tarr decidedly a less comfortable experience.

In a passage of FID added to the 1928 edition—one not present in the 1918—Bertha struggles to fit the rape into some coherent pattern, but fails: “What was this to be followed by? By nothing. Her heart sank: with the ultimate thud of nightmare it struck bottom. This was an end of all explanations.” Bertha’s concluding words might almost serve as a description of Lewis’s motivations in revising this scene. Here, with added FID and removed authorial judgments, was an end of all explanations.


Leave a comment


email (not published)