Last time in History of the Classes, I talked about a few iterations of OD&D thieves. Now we’re on to Advanced Dungeons & Dragons thieves, as well as their close kin – assassins and thief-acrobats. The changes to the Thief itself are quite scant, but I will analyze these small changes like whoa. Let’s get to it.

(Part One)

AD&D’s Thief

One of the first things the text tells us about the thief is that they are pretty useless in melee, as “they fight only slightly more effectively than magic-users.” Intrigued by this damnation-through-faint-praise, I tracked down the attack matrices of the Rules Cyclopedia and the 1e Dungeon Master’s Guide. I had never done this before, because 1e is the only edition I have never even vaguely played or run. Let me tell you, attack matrices are bullshit. I know there are people who are much better visual learners than mental-math types, and for those people I am glad that these pages exist. As the primary presentation of how to resolve attack rolls, these table lookups are just terrible, and I now understand why people had to have DM screens.

The salient point to this is that yes, thieves got slightly worse at combat; where once they had been on par with clerics and druids, now there is a new category for them, worse than clerics but better than magic-users. I’m pretty sure this wasn’t necessary, but… eh, whatever. (Note: my facts may be all out of order and wrong here. My source for OD&D was written after my source for 1e. Someone with better sources, feel free to jump in the comments, below.)

Thieves also graduate up to d6s for Hit Dice, where they will stay for many years to come – all the way up to 5th edition, in fact. Much like their attack matrix, this means “tougher than magic-users, squishier than clerics.” The huge change (from Supplement I, anyway) is that thieves can now gain a bonus to AC and ranged attacks based on Dexterity. The bonuses don’t kick in until Dex 15+, but if that’s you, it’s a great accuracy and survivability boost.

For some reason, the rest of their abilities are divided into “primary,” “secondary,” and “accrued” abilities. Gygax describes five thief skills as primary (pick pockets, open locks, find/remove traps, move silently, hide in shadows), two as secondary (hear noise and climb walls), and one as accrued (read languages – to be fair, this one doesn’t start until 4th level). Backstabbing is also a secondary ability, and here it scales up to quintuple damage at 13th level. The text is unclear as to whether it scales up to sextuple damage at 17th level; a literal but nonsensical reading indicates that thieves cannot backstab at all at 17th level and above. Given the thief’s slow attack progression, the +4 attack bonus for backstabs is a big deal.

There are two other “accrued” abilities as well: Thieves’ Cant, which Gygax calls a language even if the thing he’s talking about is an argot

Wait. If Gygax doesn’t have a way, in rules text, to differentiate between languages and argots in OD&D and 1e, Alignment Languages might really just be Alignment Argots – Common plus a few odd words sprinkled in. Still weird, but 90% less unbelievable!

…and the ability to decipher magical writings and use spell scrolls intended for magic-users or druids. The druidic scroll thing is weird; there’s a double negative on the field that fouls the whole paragraph (ten yard penalty). Glancing back at the Druid, “Druids can use those magic items not otherwise proscribed which are for all classes and those for regular clerics which are not written, i.e. books and scrolls.” Only in the DMG can I find a flat confirmation that druidic scrolls exist; it turns out that there is a 25% chance that any clerical scroll discovered is instead a druidic scroll.

They also have a 25% backfire chance rather than a 10% backfire chance. That feels a lot more like a tool of last resort than an ability you might use in a common situation.

Gygax splits a pretty fine hair when he says that thieves cannot build strongholds, and goes on to describe the strongholds they can build; for him, stronghold is a technical term that involves being far from civilization. There’s always another Master Thief in the area, and by establishing himself the PC implicitly goes to war with that Master Thief. A turf war ensues. It’s interesting, in hindsight, that the Rules Cyclopedia opened up the possibility of being one of that Master Thief’s capos and earning your own turf. Anyway, what ensues is a mini-campaign of Blades in the Dark, which is awesome but probably spotty in 1e’s implementation – in theory the PC Master Thief is the only PC involved in this mini-campaign. This would be a great place to use Ars Magica’s troupe-style play to give each player’s stronghold (or whatever) a few full sessions of focus. (Alternate summary of this paragraph: we’ve developed some awesome game-running technology in the decades since this publication.)

Finally, there’s a half-page of explanation for how each thief skill works. Some of these are a real mess – Pick Pockets is about as complicated as the tracking rules, Open Locks is “fail once and you can’t try again until you level” (which could really shut down an adventure if the DM hasn’t planned around it), Hide in Shadows is detected by detect invisibility and true seeing for some reason, and Climb Walls eventually requires a third d10 for a value to the right of the decimal. Climb Walls is especially messy because, by shifting from whole-number percentile progression to fractional percentile progression, dwarf, gnome, and halfling thieves are always stuck with a success chance somewhere in the 80s. This is nominally realistic, but bizarre when you consider how superhuman thieves are otherwise allowed to become.

The DMG presents quite a bit more commentary on the rules around thief abilities, and grants the additional ability to set traps. Here Gygax’s language becomes openly abusive toward players with the temerity to believe that the rulebook isn’t lying to them (Hide in Shadows, p. 19). It’s clear that the thieves in Gygax’s games were driving him to distraction, and he uses the DMG as a chance to slap them down. I’ve had my own fights about how to treat stealth in other editions of D&D, so I’m not completely without sympathy, but it is a terrible idea to have a chapter of rules about how PCs’ abilities secretly aren’t as good as you’ve led them to believe. This is a source of decades of player-DM mistrust, to say nothing of game-stopping arguments.

Anyway, that’s pretty much thieves for you in 1e. They’re nearly identical to their OD&D predecessors, with some retuning of success and failure chances for various abilities, and an average of one more hit point per level.



Let me start by quoting what must be the most perfectly Gygaxian sentence ever penned:

“Assassins are evil in alignment (perforce, as the killing of humans and other intelligent life forms for the purpose of profit is basically held to be the antithesis of weal).

All of the 1e classes suffer from idiosyncratic information presentation, but maybe none more so than the assassin. Other than the ability score minimums for assassins – and I have no idea why this class is more demanding than the thief – the first rules text of the writeup is a buried reference to their ability to use any weapon and a shield. I’ve never understood why D&D assassins can use shields, but it shows up again in 4e.

Assassins can use poison, and here we have a bunch of one-off rules on poison detection and how people react that might not apply for non-assassins using poison. That outcome is ridiculous, of course – my point is that this is a terrible place to store the information. They also have a chance for an instant kill on a target they surprise – an ability that survives in various forms all the way to 5e, though it’s only thematically present in 4e. In 1e it’s a percent chance of success, which the Player’s Handbook lists as “roughly 50%.” This value is generally about right, per the DMG, if your intended victim is the same level as you, but of course it’s a chart lookup that is also the percent chance of a coup de grace on a helpless target.

Assassins also use disguises extensively, a theme that 5e picks up again. In this vein, they learn Alignment Languages that are not their own, including Druidic and Thieves’ Cant. They can also pass themselves off as a different race, class, or gender, but with a percentage chance per day of being detected. This percentage is awfully complicated to calculate, and if you understand math at all, you start to realize that failure is all but guaranteed if you have to stay in disguise in front of more than a tiny number of observers for more than 24 hours. Every observer rolls their percentage separately, and if you have to check ten times, there starts to be a pretty good chance that one of those rolls comes up 08 or below. Oh, and detect evil – the at-will paladin ability – is sufficient to foil this. In short, you’re a low-rent Faceless Man; where Faceless Men are overtly magical in a very-low-magic setting, assassins are nonmagical in an omnipresent-magic setting.

Assassins have the option to spy rather than murder. Um, good to know?

For all thief skills other than backstabbing, assassins have the percentages of a thief two levels lower than the assassin. This is fine and good, though it makes it really weird that the Assassin class has a hard cap of 15th level, while most classes can advance without explicit limit (Druids and Monks are the other big exceptions here).

There’s also a chart describing expected fees (what’s the point of having a Guild if you can’t engage in price-fixing?) for assassinations. Because it’s 1e, you get paid twice in XP – once for the victim according to his or her level, and once for the money you receive. I guess that explains why assassins don’t receive an XP bump for high ability scores – they are the only class that earns more XP from solo, non-adventuring actions than from normal gameplay (I’m guessing, but the rewards are quite nice).

The rest of the Player’s Handbook description of assassins deals with high-level play and advancement in Assassins’ Guilds. The main issue is that, much like high-level druids and monks, assassins have to eliminate the next guy up the ladder to advance. Here at least it makes sense that a class about death and criminality has to commit murder, but conflating social role and class level can’t help but have weird unintended outcomes. Even weirder, any newcomer to the Grandfather of Assassins position is (for whatever reason) obligated to destroy the previous headquarters, pay exorbitant amounts of money to all lower-ranking assassins, and construct a new headquarters. I expect this would be quite a shock to the Assassins of Alamut Castle – even though all of Gygax’s descriptions of the Grandfather’s headquarters point to Alamut and the Old Man of the Mountain.

There’s one other really interesting line here. When a new Guildmaster claims a territory, 75% of the assassins in that area leave. “Thus, it will be necessary for the new Guildmaster to allow new members into the guild. These new assassins will all be 1st level and must be worked up in experience levels.” The sheer weight of the downtime mini-campaign implied by actively building up your followers’ levels is unbelievable; the bookkeeping of a Birthright campaign looks trivial.




…for people currently playing assassins in 1e campaigns, or expecting to do so…




Oh, did you think we were done? Sorry, there are also Top Secret DMG rules. Assassins can set traps just like thieves, and there is a super secret class feature to research more about poisons at or after 9th level. The book goes out of its way to say that the DM should hide this class ability from the player… as if players are in the habit of asking about things that might be unlisted class features with half-page-long subsystems. Basically, this class feature is for people who have made an exhaustive study of the DMG before deciding to play an assassin. It amounts to a handwave at a research and crafting system for poisons, and if you don’t do this, you’re substantially worse at using poisons – literally the first major class feature in the Player’s Handbook writeup.

The takeaway here is that game design and information presentation have grown a lot since 1979. Beyond that, the assassin is a reasonably interesting class that would be a lot more so with some sort of creed, or a way to use their class features to better ends. Vlad Taltos is easily one of my favorite characters in fiction, though, and I’m deeply biased.


Unearthed Arcana: The Thief-Acrobat

The thief-acrobat isn’t exactly a class unto itself. In fact, it is very close to being a 5e-style subclass that you select (instead of “thief”) at 6th level. I recommend this highly if you want to watch your DM’s eyes skeet blood. This subclass has a lot going on. The good thing about it is that you can choose to be a second-story man and cat burglar in preference to being a box-man or pickpocket, and that’s a nice level of customization. The bad thing is that you have four separate abilities that do pretty close to the same thing, The bad side is that every attack (apparently including small area-effect spells) against you that comes after you in initiative order suffers an evasion chance in addition to needing to hit you or make you fail a saving throw.

The thief-acrobat gets:

  • No further advancement in Pick Pockets, Open Locks, Read Magic (she never learns this ability), or Find and Remove Traps.
  • A 10% bonus to XP if they have Str 16 and Dex 16.
    • The minimum ability scores to be a thief-acrobat: Str 15, Dex 16. Sort of an odd inclusion.
  • Tightrope-walking, a percentage skill that starts at 75% and climbs sharply. There are restrictive load limits, measured in gold coins, and penalties for wind conditions. At higher levels you can carry more weight and you reduce the penalties.
    • The system approaches limitations and penalties differently with every single percentage skill, from Pick Pockets (which goes above 100%) to Climb Walls (which gets asymptotically close to 100%) to Tightrope Walking. Consistency is all we seek!
  • Pole vaulting – it is not a percentage skill, just a height that the thief-acrobat consistently clears.
    • The text mentions that “Non-thief-acrobats cannot effectively pole vault.” Do you remember last week, when I mentioned that giving the thief these skills meant that fighting-men, clerics, and magic-users implicitly lost the potential to attempt them? This is an explicit case of exactly that.
  • Jumping, with stats for high jumps and broad jumps. The jumping rules for all other classes are, obviously, stored here as well. Because you would ever think to look here for that rule.
    • The world-record high jump, made by Cuban athlete Javier Sotomayor in 1993, is on par with an 18th-level thief-acrobat (8 ft ½ in). When Unearthed Arcana was released, the record-holder was Vladimir Yashchenko of the USSR (7 ft 8 ¼ in), or a 17th-level thief-acrobat. You heard it here first, folks: thief-acrobats are all Commies.
    • For what it’s worth, the top distance for 23rd-level thief-acrobats making a running jump is more than seven feet short of the record at the time UA was published.
  • Tumbling, which has three distinct options: attack, evasion, and falling.
    • Attack is a bonus to pummeling attacks. In case you’ve never looked up pummeling in the 1e DMG, you really should. It drops all of this d20-rolling for a far more complicated system that makes an illegitimate love-child of Synnibar and RIFTS look pleasantly straightforward. I am looking at it right now, and I have no idea what in God’s name it says.
      • In other times and places, I have talked about the excessive complexity of tracking systems, grappling systems, and so on. You guys, I had no idea. This is the most pointlessly complicated system I have ever seen attached to D&D, and I read all of the optional rules around spellfire in 2e Forgotten Realms.
    • Evasion is a percentage chance to dodge an attack or small area-effect out of hand, before even considering Armor Class or saving throws. It’s not clear in the text whether this takes your action for the round or not.
      • This is interesting, because before right now I had not realized 3.x’s Evasion rogue feature had an antecedent.
    • Falling is a percentage chance to ignore or reduce falling damage. It is, once again, very complicated to resolve, but it’s here that we see that Gygax intended falling damage to be much greater than “1d6 per 10’ fallen” implies. (I realize many people have reported this before in the blogosphere, but bear with me.) It’s 1d6 for the first 10 ft, 2d6 for the second 10 ft (so falling 20 ft is 3d6 damage), and so on.

There are a few more charts to give thief-acrobats more benefits for high ability scores, followed by a special encumbrance subsystem that applies only to thief-acrobats. It introduces the concept of body-associated weight, as opposed to carried weight. You know what, I’m gonna let it go.

The thief-acrobat is a cool character in principle, but the mechanics are a mess, and I’m guessing that most campaigns don’t have nearly enough places for tightrope-walking, pole vaulting, and daring leaps that only one character needs to make. In short, this is the parkour class – if you could multiclass it with assassin, it would be Daniel Craig’s James Bond. The other thing about this class is that a surprisingly large amount of its DNA gets woven back into the 2e Thief to create the Rogue of 3.0. That’s the kind of design move that gives my column a reason to exist!