First, of all, it looks as if Google digitalised the long s (ſ) as s most of the time.
Then, one has to consider the following four variants of s-spelling. I distinguish between the case where a long s (ſ) is used (mostly blackletter fonts) and where it is not (mostly antiqua fonts):
- Heyse’s rules – ſs (ss) after short vowels:
dass, müsst, ließ (antiqua);
daſs, müſst, ließ (blackletter)
- »modified Heyse« – ſs (ss) after short vowels, ſſt instead of ſst:
dass, müsst, ließ (antiqua);
daſs, müſſt, ließ (blackletter)
- Adelung’s rules – ß also after short vowels:
daß, müßt, ließ (antiqua and blackletter)
- No ß at all (Switzerland and fallback solution, if ß is unavailable):
dass, müsst, liess (antiqua);
no blackletter equivalent (as far as I know)
Since muss and dass are affected equally by all four rule sets, whoever spells dass will almost certainly spell muss and not muß. This mostly answers question 4. In addition, muss and dass are almost certainly the two most common words in which the four variants differ. Therefore they are less affected by statistical fluctuation and most writers were confident about the spelling of those words. For rarer words like Fluss and Kuss, the similarity is less prominent. The similarity is also smaller for words, for which variant 1 and 2 differ. Since they should not differ, if ſ were always digitalised as s, I assume that ſs was digitalised as ß sometimes.
A good way to find out, what is going on, is to separate the case, starting with variant 4, since it can be extracted most easily by looking at words like Straße that are written with ß in all other variants. As can be seen from this Ngram, variant 4 accounts almost perfectly for the remaining dass spellings between 1910 and 1996. I would explain its detailled temporal evolution as follows:
- From 1901 to ca. 1925, variant 4 is strongly decreasing. This is probably due to the antiqua ß becoming widely available (it was only »invented« in the 19th century), thus reducing the need for the fallback solution ss. Another factor might be people who refused to use the ß altogether (at least in Antiqua) and only started to do so as the official rules became widely accepted (see below).
- After this, variant 4 is gaining popularity again slowly until ca. 1940, which is most probably due to the ß getting out of use in Switzerland.
- There is a strong peak for variant 4 in 1945, which might be due to World War II, which can be assumed to reduce the text output by the fighting countries. Since Switzerland did not participate in the war, its output can however be assumed to be rather constant at the same time, which would raise the relative impact of Swiss books and thus of variant 4, explaining the peak. Afterwards, the book output of the rest of the German speaking world rose again, thus reducing the relative impact of Swiss books and variant 4 to normal. (This assumption is also backed by this Ngram, if you assume that Swiss books use words referring to Switzerland more often.)
- Afterwards the fraction of books that uses variant 4 remains mainly constant, since Switzerland had almost fully abolished the ß by then, while elsewhere, the antiqua ß as well as the orthography from 1901 were sufficiently established, such that almost nobody needed to use the fallback solution anymore.
Now, let’s look at the remaining time:
Before 1901, things were complicated. There were no generally accepted official spelling rules, however certain governments or parts of them issued spelling rules, e.g., certain schools in Austria were directed to use Heyse’s rules in 1879. As can be seen from this Ngram, it looks as if variant 4 was becoming increasingly popular – however this might also be due to antiqua fonts becoming more popular and people not using the ß in these fonts, e.g., due to uncertainty about spelling, lack of the glyph, etc.¹ However, when subtracting the ratio of people who presumably used no ß at all, we obtain an estimate of the fraction of users who used variant 1 or 2. Only between ca. 1890 and ca. 1910 this estimate seems to be more than just statistical fluctuation, and then it is about 5 %, while variant 3 is used by about 35 %. So, when variant 3 (Adelung) was established as part of the new official spelling rules in 1901, it can be estimated that it was already used by the majority of those who used the ß at all. The sharp drop of dass and muss at that time must thus be attributed to other factors, e.g., people starting to use the ß at all in certain fonts¹.
Finally, the orthography reform of 1996 reintroduced Heyse’s rules, which explains the sharp increase of dass and muss after that year.
¹ Note that the ß originated from blackletter fonts and existed in those for a long time. It can therefore be safely assumed, that blackletter fonts without ß were not an issue – moreover since most blackletter text printed worldwide at that time was German.