Benches in first line position: some progress

A little while ago I made a post about benches [ch] and [sh] at the start of a line. In the post I attempted to “restore” words with initial benches by removing preceding glyphs. The attempt wasn’t a great success and I could not find the solid answer that I wanted.

I’ve begun approaching the problem of line patterns more systematically, which I hope will provide more answers over time. One part of that approach is to work with parts of the text with a single scribe and topic. So everything in the following post will apply only to text written by Hand 1 on herbal pages.


Words with initial benches are less common in that first position than in the rest of the line. Yet the total number of word containing benches remains broadly unchanged. The table below provides the relative frequency of words containing [ch] and words with initial [ch].

Containing [ch]0.410.420.390.370.32
Initial [ch*]0.050.300.250.220.19

I have limited the examples to [ch] as the pattern is stronger and clearer than for [sh]. (The frequency declines to the right due to another line pattern which reduces frequency in the last line position.)

The total number of words with initial [ch] which are missing from the first position is around 250-300 words, but it is impossible to estimate precisely.

What remains

In the earlier post I looked at words which might contain the benches which had been “moved” to the middle of a word. It appears that the glyphs [y], [d], [o], [s], [t], and [p] all have a higher frequency before [ch] in the first line position. But I don’t want to discuss that yet.

The more interesting question concerns the words with initial [ch] which remain in the first line position. If we look more closely at them we can see a striking pattern. (The table below shows the number of occurrences for words beginning with the specified strings, including those for which the string constitutes the complete word.)


The numbers are very clear: almost the only words with an initial [ch] in the first line position are those beginning [cho*]. Words beginning [che*], for which we might expect to see ten occurrences even at the reduced rate of initial [ch], simply don’t occur.

We can go one step further with this line of investigation. Words beginning [cho*] are themselves less common than expected, even if they do occur. Is there a pattern to what words with initial [cho*] remain?

(As before, the table below shows the number of occurrences for words beginning with the specified strings, including those for which the string constitutes the complete word.)

[cho] (Complete word)2161383

We can see that while words beginning [chok*] and [chot*] remain in line with the expected numbers, all other words are less common than expected. Words beginning [chor*] and [chol*] are particularly less common (most of these, in all positions are the complete words [chor] and [chol]).

(As a side note, words with final [l] are less common in the first line position, but words with final [r] are more common, so the numbers about about [chor] and [chol] don’t clearly fit into a wider pattern.)


The process which decreases the number of words with initial [ch] in the first line position is not random. It does not indiscriminately reduce the numbers of such words so that they are all less common. It strongly affects some words but leaves others untouched.

Words beginning [che*] and [cha*] are all gone. The words [chor], [chol], [chy] and [cho] are significantly reduced. But words beginning [chot*] and [chok*] are unaffected.

Of those words beginning [chot*] and [chok*], only three have more than one token in the first line position, the rest being unique to this position. The process is not leaving specific words untouched, but words with specific features. The process is not dumb, and it does not work strictly on the initial glyph alone.

For those words with initial [ch] which appear to gain an extraneous glyph to begin [dch], [ych], [och], [sch], [tch], and [pch] in the first line position, it remains unknown why one glyph is chosen over another. It has been proposed, quite reasonably, that there is an influence from the last word of the line before. Yet the information in this post suggests that the process is aware of, and responsive to, the features in the word it affects. Words beginning [dch] may have some feature in common with one another which those beginning [ych] or [och] do not.


9 thoughts on “Benches in first line position: some progress

  1. Fascinating post, as usual! But I have a question.

    You write “Yet the information in this post suggests that the process [of prefixing other characters to line-initial [ch]] is aware of, and responsive to, the features in the word it affects. Words beginning [dch] may have some feature in common with one another which those beginning [ych] or [och] do not.” It seems to me that the data in this post supports the first conclusion but not necessarily the second: there is nothing here about the choice of frequency of prefix characters.

    Am I wrong? Did I miss something? Did you mean this in conjunction with another post on this topic?


    • Hello Marnen, maybe I have expressed my thoughts ambiguously. As always, it is hard for us to catch the ambiguity in our expressions as we know how they should be read. The alternatives don’t occur to us. I suppose I should restate what I mean, but more fully to make it clear.

      1. There is a process which adds extraneous glyphs to the beginning of words with initial [ch] when they occur at the beginning of a line.
      2. Some words with initial [ch], such as those beginning [chok*] and [chot*] never receive extraneous glyphs.
      3. The process in (1) is thus aware of other features in the words it works upon, beyond the initial [ch].
      4. The process in (1) has the potential to add a variety of extraneous glyphs to word with initial [ch], including [d], [y], [o], [s], [t], and [p] (although the last two may be part of another process).
      5. Given point (3) that the process is aware of other features in a word, there is the possibility that the glyph chosen from those listed in (4) is dependent on a particular feature present in the word it attaches to.
      6. Following from (5), we might assume that words beginning [dch] share a specific feature, and that words beginning [ych] share another feature, and that [och] share another feature, and so on.
      7. [Implied but not stated.] Searching for the specific features which cause [dch], [ych, [och], and [sch] would be a worthwhile route of investigation.

      I think the ambiguity comes at point 6: I was making an assumption about the possible, not a statement of what has been proven. Does that seems clearer?


    • I hope they’re useful! Sometimes I’m trawling through stats and spot something and feel the need to tell others.

      In this case, I suppose I simply wanted to say, “this process isn’t random, it’s aware and specific.” While I can’t claim all line patterns are the same, it does at least encourage us to look for more rules for how they work.

      I expect it will take a dozen posts before we even understand one line pattern, never mind the lot!


      • I strongly suspect that you will find similar specific patterns for just about every character that appears at line-starts. 🙂

        My own favourite line-start mystery glyph (or pair) is EVA s: in many places where line-initial s appears, I can’t help but think that it feels more like a null than a genuine letter. All the same, I haven’t really got a good plan for how that ‘feel’ could be properly communicated / demonstrated / proved. Just more wood to throw on the same line-initial fire. 😉


  2. Hi Emma, I recently found that first-word statistics (in general) are considerably skewed by the recipes section in quire 20. What happens there is that every second or third line is the start of a new paragraph. Because of that, the ‘first line in paragraph’ effect gets mixed into the other stats that one might be interested in. I didn’t check the specific case that you are testing here, but suspect that there would be some impact.
    By separating ‘first lines in paragraphs’ from ‘all other lines’, one can get separate statistics.
    In my case, I found that the observation that first words in paragraph text are on average longer than the others, is almost entirely due to this.


    • Hi Rene, I did some work on the Quire 20 first word statistics a few years ago, and I think I ran into this problem too. So many Grove words!

      The problem I have about separating of the first words of paragraphs is that it makes it harder to know where the “missing” words went. For example, I need to know how many words begin [pch] so I can then add them into the figures to account when looking at words with initial [ch].


  3. Yes, that’s exactly the point. It is likely that the vast majority of [pch] are on first lines of paragraphs. Now if there were always (say) 10 lines per paragraph, then the ‘initial-line issue’ would not have a great impact on the statistics, but in Q20 it is very different.
    Effectively, the table:
    Initial [ch*] 0.05 0.30 0.25 0.22 0.19
    Could look very different for ‘first lines’ and for ‘all other lines’.


    • I’m sorry René, I don’t understand your point. I’m probably making an assumption about how line patterns work which stops me from understanding. My assumption is that while the initial [p] is always at the start of a paragraph, any glyph which follows is random. So the appearance of [ch] is not tied to [pch]. Does that make any sense?

      (Although, the glyph following an initial gallows at the start of a paragraph isn’t actually random. It seems as though an extra [y] or [o] is inserted if the word begins with certain glyphs. Haven’t done the research to prove that though.)


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s