Brains, Looks, Credibility*, Sex, Health, Strength, CourageThe so-called "Dalluhn Manuscript" or "Beyond This Point Be Dragons" document, the following statistics appear on the character write-up (you can see it in a picture here):
Horsemanship, Woodsmanship, Leadership*, Flying, Seamanship, Cunning
* these are spelled poorly but you get the idea
Intelligence, Cunning, Strength, Health, Appearance, EgoFrom the Gaylord sheet to Dalluhn (likely a playtest document or a copy of one), we see that things are a bit more formalized, with Brains clearly becoming Intelligence, and Looks going to Appearance. Credibility, Courage, Sex, Leadership, and the skill-like values (Horsemanship, Woodsmanship, Flying, Seamanship) are gone entirely.
In the Gaylord sheet, these are in the range from 3 (Gaylord's Seamanship) to 11 (his Brains, Flying and Cunning). Several values are crossed out and replaced with increased values. It's not clear whether these values were increased through play, or if they were changed for some other reason, but it seems entirely within reason that attributes would increase in Arnesonian D&D; it was very much Gygax's bailiwick to make them extremely static. Against this, many other games of the very early days of RPGs, such as Tunnels & Trolls and Runequest, used very similar statistics but assumed that they would be increasing as the character's adventuring career went on.
Over at the Comeback Inn, Greg Svenson says, "In my recollection, the stats were used more for task resolution than for adjustments in combat." This squares with Arneson's tendency to have players roll 2d6 for a great deal of task resolution: Gaylord's original Strength was 5, so on a Strength related roll he might have had to roll a 5 or less on 2d6. This same logic works in OD&D, where ability numbers have very little impact on the probabilities of combat, but many referees have used them for roll-under task resolution, such as calling for 3d6 or 1d20 under a specific stat to resolve an action not documented in the rules (which in OD&D is most non-combat actions).
Arneson was using this as a work continuously in progress. For instance, in the Comeback Inn link, Svenson mentions that the "Sex" stat was not, as may be expected, particularly involved in the more Gorean parts of Arneson's game. The "standard six" of D&D as codified by Gygax, even in the correct OD&D order (Constitution before Dexterity), are tried and true but far from exhaustive. Arneson was much more freewheeling, and felt free to define things that we might call "skills" but he treated just the same as any other characteristic. There is something to be learned from that attitude.
One of the implied uses of these statistics was for saving throws, which makes a neat parallel for later games like Castles & Crusades and 5th edition Dungeons & Dragons. It's doubly interesting that "Dexterity" – a favorite for stat-based saving throws involving dodges – appears neither in the Gaylord sheet nor in Dalluhn, even in some form that's not a word dragged out of a thesaurus by Gary Gygax. Most of the modern "Dexterity" score is actually a combination of dexterity (hand-eye coordination / fine motor tasks) and agility (quickness of reaction and ability to move easily). As an oddity of history, Daniel Boggs's Arnesonian reconstruction, Dragons at Dawn, wound up stuck with a very out-of-place Dexterity.
I also like a few of Arneson's stats, particularly "Cunning." Think of the difference between a game with a Wisdom stat and one with a score for Cunning; it's like the difference between a Tolkien novel and a Fritz Leiber short story. (I think this is why, for instance, Dungeon Crawl Classics uses Luck over Wisdom.) Appearance/Looks and Credibility are also an interesting twist on Charisma, although I'm less keen on "Sex" given the looks of many gamers. I'm really thinking that tinkering with ability scores is an easy way to give your D&D a small Arnesonian twist.