The next manuscript I’ve reviewed is the Paris graecus $2391$, a copy of which can be found here.
History
Peters & Knobel give virtually no background on this text, only noting it originates in the $15^{th}$ century.
Handwriting
This manuscript is quite lovely. The handwriting is impeccable and very easy to read. The characters, by and large, conform to the Greek uncial from the Paris graecus $2389$ so I’ll save myself the effort of going through each character here again save to mention the variations.
The first is the symbol for $0;40$ which is similar to a “w” but with a curl on the end and a dash over that.
There are a few characters which take various forms.
As with the Paris graecus $2390$ manuscript, we find beta ($2$) written both as β and the cursive form that looks like the one from the $2390$ manuscript: 
. We also find delta ($4$) both as d and δ as well as epsilon ($5$) both as e and a cursive form that looks like a G. There is also a variant on the character for $7$. The most common form seems to be a fairly standard ζ. However, the cursive form seems to be almost turned $180º$.
Ultimately, the characters in this manuscript are generally easy to distinguish. The largest difficulties lie in the differences between the characters for $40$ and $50$ and occasionally between $8$ and $20$. Errors and corrections were similarly rare making this manuscript one of my favorites to read thus far.
Additional Notes
One of the immediate oddities is that some of the data appears to be missing on first inspection. The columns are clearly present for the sign and the direction north/south of the ecliptic, but appear to be empty.
However, values can very faintly be seen1 and I was able to bring out by selecting just these columns and adjusting the brightness and contrast.
My suspicion is that these values were written with a different type of ink that faded over time. I suspect it may have been a different color from the rest of the text as I also note that there are some places in the descriptions where faint text is also visible. These appear to be in places where I expect the constellation headers to be and would be an excellent place to make use of a different colored ink to draw attention to them. Having seen some other manuscripts, I suspect it may have been a red color. Talking to a friend who does scribal work, they suspect it may have been a madder ink but I have no way to confirm this.
As far as the values provided by Peters & Knobel, I again find frequent errors in which variants from Baily are not correctly reported. Indeed, errors seem to be more common than in the previous manuscript2. Here, they average at least one per page.
This manuscript lacks the dramatic flourish on the last character but there is still a hit of the scribe’s mindset that comes through as suddenly, in the last few pages, they likely realized they were running low on room and suddenly started squeezing in four lines per division instead of the typical three.