Also, I put most of my legendary military dwarves on scholar duty.
Either observer is an academic skill too, or their name comes from having a high Teacher skill, but none to dabbling in all other academic skills.
I think the latter.
Their academic profession is called 'naturalist'.
It's not the tracking skill, even though they are all dabblers there, because my non-military hunter, who is better at tracking, but lacks the observer and teaching skills, is just a 'scholar'.
In fort I've noticed that dwarves will not gain a scholar profession until they gain a level in it. I will go through my forts and see if I can prove observer as a scholar skill. I haven't been checking for it because I didn't think of it. I have found no correlation between the teacher skill and any academic profession title but I think I had a migrant peasant with a high teacher skill and nothing else a few forts back so I will check. Since I am looking at observer I will also look at other common military skills to see if I can find something.
Back on task, according to the wiki most jobs give either 30 or 10 experience. 95% of jobs give the standard 30, while mass jobs (like stone detailing or firing at an archery target) give just 10. However, attacking a living target gives a huge experience boost for weapon skills. Due to the extreme length of time it takes scholars to finish their various tasks, they rarely get very far trying to train in-fortress. I've seen cheesemakers with great intellects that valued knowledge spend years without breaking "dabbling" in their skills new scholarly skills. They even gained masters, but the master didn't know much crap in the first place >.> Skill gains from master/student relationships should have a higher modifier than just the base 30. You are getting direct 1v1 instruction from someone you respect enough to call "master," you should be learning far more than at a normal rate researching on your own or just arguing with random Urists. Even then, it'd still be slower growth. I don't expect them to pop straight to legendary like a miner tearing through a sand bank for several months, but spending a week on the same job to only gain 10 exp is too much in the other direction.
Skill growth for scholarly activities appears to be split between skills. 1 discussion for example will give 10-30 in speaking and 10-30 in chemist (or w/e topic). Same for book writing which will give a (I think) standard 6 in wordsmith, 12 in writing, and 10 in a topic skill.
I do agree that skill growth is slow but I'm working on a way to train them faster. I think its all about the library set up. Previously in .02-.03 forts I saw skill growth all over. However, since starting .04 I have not seen dwarves discussing topics that they do not know about. My current fort has 12 scholars, all with medical skills except 1 that has tracking and mechanics. For the past 5 years the dwarves have only talked about these topics. In those 5 years nearly all of them have gained at least 1 skill level in each. My idea here is based on the idiom "you don't know what you don't know" so my next test will be as follows;
Embark with 3 scholar dwarves (the other 4 will build the fort so I don't contaminate my scholars with job skills), all will have a 3 in writing and 2 in teaching with the following skills
Dwarf 1: 5 Diagnose
Dwarf 2: 5 Geography
Dwarf 3: 5 Chemist
I will then set up 3 different libraries who have no scholar visiters and test what skills are gained when. I will have to be careful as to what skills any new scholars have that I add to the libraries so as not to contaminate my experiment. If I get different topic skills from migrants then I will open up new libraries.
Ideally I want at least 5 scholars per library to maximize the amount of activities undertaken. I hope unskilled dwarves don't mess it up, I may have to savescum a few times to preserve my data.
My hypothesis is that if we decrease the amount of potential topics then we will see an increase in skill growth in the remaining topics.
Now I did a similar test in .02 but like I said dwarves appear to not discuss unknown topics in .04.