You are asking all the questions I asked, I am not an expert in this, just someone that has travelled your path before.
The reason you wind more turns is so that you don't have to add turns. It is adding turns that is problematic once you have "rubbed off the enamel". You don't have to rub off the enamel lots, in fact just once if you are careful, because removing the turns doesn't break the wire, and if the wire is straight not wound it doesn't create measurable inductance. That is to say you can hook up an inductance meter to the wire, and the inductance will fall as you unwind. With ferrite cores you can snap the core on and off pretty easily (not so easy with toroids).
+1 to the idea of wind 100 turns and see where you are.
How to: rubbing off the enamel
1) much of the wire you will find is designed so that high heat soldering turns the enamel into flux (bad flux, but flux sort of ) but 2) razorblade scraping wire on a piece of wood does the trick for any enamel.
Audibility of saturation.
The shape of the distortion resulting from saturation is different based upon material, but generally you are attempting NOT to saturate the coil. The topic of inductance and what materials have what AL is something of a black art, with real work started by Bell Labs I think, is interesting, and was pretty much wild experimentation with materials and how to process them (sequence of heat and cooling matters ... lots of trial and error. Interesting reading, not particularly illuminating for building an EQ except the "boy I am glad bell labs did that so I don't have to" feeling you get!.)
Pultec originally used a toroidal core (hard to wind, can be done by hand but dedicate days if you are like me) I believe the common wisdom is that they used a light grey colored toroidal core the material is though to be MPP (this is powedered material sort of sintered (look it up) into a core material. The sintering creates tiny gaps in the material and keeps the AL value lower and saturation levels higher. MPP is supposed to saturate more gradually than ferrite. ). I have produced copies of pultec inductors in the approximate size and material as best I can and all I can say confidently that they were no where near saturating the core in the Pultec circuit.
So you want a material that will work for your values in audio
Generally you are looking for a gapped core with an AL of around 400. AL of 630 is fine. Ungapped core will have AL values so high that you will take very few turns to do the job, small number of turns will saturate more.
I did some experimentation using various materials, on a core with "taps" at 21mH to 175mH approximately (about what is used in the Pultec eq). I found that I could saturate the coil will Audio levels at 20hZ at the LOWEST TAPS on the UNGAPPED core. I was testing with an Oscilloscope not audibly. I was measuring the RMS voltage needed to saturate the core (there is a thread here on how to do that search ferrite audio core saturation or something, let me know if you can't find it. Use CJ's initials (Mr Inductor... with an Axe).)
I have data on this if you are interested, but I could only saturate the gapped cores at about 20dB, which is a level never seen in the attenuated levels INSIDE an EQ (before the make up gain).
The amount of wire that fits on a bobbin is calculated using what is called "circle stacking" and you can look up that term and "reel factor", you calculated the depth of the bobbin reel, and the width of the bobbin reel, and the amount of space you want to leave and that gives you an area and that area has "circles" stacked in it (cross section of the wire is a circle). One circle per turn. I have a spreadsheet for this.
Here is some wire data, in inches, with wire gauge, diameter with enamel, resistance per foot.
Wire Gauge Thick w/coating Resistance / Foot
22 0.0281 0.01614
26 0.0182 0.04081
30 0.0121 0.1032
32 0.0097 0.1641
34 0.0078 0.2609
36 0.0063 0.4148
38 0.0051 0.6596
40 0.0041 1.049
42 0.0033 1.635
44 0.0025 2.554
Here are some other useful statistics, I have a spreadsheet that calculates all of this but it is a bit complicated and you might just want to do your own.
How many turns fit on a bobbin (this is not the ideal number assuming a perfect circle stack, this is what I can achieve (barely) winding carefully with a drill)
RM8
(Gauge) (Turns)
32 463
34 716
36 1,097
38 1,674
40 2,591
You can calculate the Q factor by using resistance per foot and the length of the wire
RM8 Core turn lengths in inches
Shortest turn Longest Turn Average length of a turn
1.22836169 1.94464421 1.58650295
Using the AL factor you can calculate a number of turns for each tap. And you can do the trial and error method. But I suggest that you do it more simply, because you are using RM8.
1) Make a "test bobbin" with 100 turns of wire on it.
2) Assemble the core.
3) measure the inductance in some way (meter, other method)
4) Remove the bobbin and keep it to do this next time.
5) You now have an EXACT AL for that EXACT CORE (each core varies).
6) Plug that AL into your spreadsheet and wind the exact number of turns it specifies on a different bobbin.
7) repeat
The result is easily within 5%, and if you have that little trimmer screw you can get it much closer (the screws cost like $2 a piece, require a core with the hole, and are not really needed). If you look at the frequency effect of 5% differences in inductance you will realize that you don't need it within 5%, never mind closer.
If you want the spreadsheet i use, it is a google sheet, and I am happy to share, but I need an email so PM me.
In general on the "how many henries will fit" category
At AL 400 with 36 gauge wire on an Pot Core 26 x 16 core you will get 420mH using 1024 turns and your DCR for that coil will be 70 ohms or so.
At AL 630 with 32 gauge wire on an RM8 core you will get your desired 35mH using 235 turns and your DCR will be 5.11 6.3 ohms (DCR will be less than that actually because that is averaging a full core and your turns will be shorter with a half full core). Correction, I was using a Pot Core 26x16 bobbin size