A few plots based on the Zandbergen-Landini transliteration (ZL_ivtff_2b.txt) and color-coded according to the scribes identified by Lisa Fagin Davis (You are not allowed to view links. Register or Login to view.). As always, it is possible I made errors along the way.
This tries to replicate the plot You are not allowed to view links. Register or Login to view.: the Y axis shows (k-t)/(k+t) while the X axis correspond to the order of pages in the manuscript.
As Nablator pointed out, there is not much correlation: I think I can see a light rising trend left-to-right, but this does not seem to depend on scribes. In particular, HerbalA and Pharma (both by Scribe 1) appear to be well distinct: slight prevalence of 't' in HerbalA and stronger prevalence of 'k' in Pharma.
This plot show the percentage of t (X axis) and k (Y axis) with respect to the total of EVA characters in each page (all spaces were ignored). Again, there does not seem to be any clear separation between scribes. Here an anti-correlated trend is quite clear: when there are many 't's there tend to be fewer 'k's and vice-versa.
This third plot shows the longest 't' sequence (X) and the longest 'k' sequence (Y) for each page. These numbers are not robust with respect to transliteration: e.g. Takahashi misread a couple of occurrences of You are not allowed to view links. Register or Login to view., breaking one of the longest k-sequences. Similarly, the You are not allowed to view links. Register or Login to view. depends on how the weird ligature at the start of the last paragraph is rendered.
It seems interesting that the outliers with high values were produced by four different scribes. Apparently, Scribe 2 took part in both the longest k-sequences (f83v) and t-sequences (f86v3 and 4).
I would like to compare these figures with actual linguistic texts. We know that these sequences are different from a randomly arranged text, but I am curious to see if they are as incompatible with natural languages as I expect.
This tries to replicate the plot You are not allowed to view links. Register or Login to view.: the Y axis shows (k-t)/(k+t) while the X axis correspond to the order of pages in the manuscript.
As Nablator pointed out, there is not much correlation: I think I can see a light rising trend left-to-right, but this does not seem to depend on scribes. In particular, HerbalA and Pharma (both by Scribe 1) appear to be well distinct: slight prevalence of 't' in HerbalA and stronger prevalence of 'k' in Pharma.
This plot show the percentage of t (X axis) and k (Y axis) with respect to the total of EVA characters in each page (all spaces were ignored). Again, there does not seem to be any clear separation between scribes. Here an anti-correlated trend is quite clear: when there are many 't's there tend to be fewer 'k's and vice-versa.
This third plot shows the longest 't' sequence (X) and the longest 'k' sequence (Y) for each page. These numbers are not robust with respect to transliteration: e.g. Takahashi misread a couple of occurrences of You are not allowed to view links. Register or Login to view., breaking one of the longest k-sequences. Similarly, the You are not allowed to view links. Register or Login to view. depends on how the weird ligature at the start of the last paragraph is rendered.
It seems interesting that the outliers with high values were produced by four different scribes. Apparently, Scribe 2 took part in both the longest k-sequences (f83v) and t-sequences (f86v3 and 4).
I would like to compare these figures with actual linguistic texts. We know that these sequences are different from a randomly arranged text, but I am curious to see if they are as incompatible with natural languages as I expect.
![[+]](https://www.voynich.ninja/images/collapse_collapsed.png)