AD&D's Most Expensive and Irritating NPC

No NPC embodies the Gygaxian maxim that dealing with NPCs "should be expensive and irritating" quite like the sage.

Like the spy, the sage is a special class of expert hireling which follows its own bespoke rules and procedures. Also like the spy, their main gameplay function is the recovery of information on behalf of the player characters. They are the "encyclopedias" and "computers" of the AD&D milieu:

Each sage has a chosen field of study (their major field), expertise in two to four special categories within that field, and some knowledge of one or two other fields of study (their minor fields):

Intuitively, the higher one rolls on the above table, the better the sage, with a roll of 91 to 100 indicating a sage with knowledge of four special categories in their major field of study as well as knowledge of two additional minor fields of study.

A sage's major and minor fields of study are determined by rolling on the table below:

Let's say I roll d100 and get a result of 76, which is a sage with knowledge of three special categories in their major field and knowledge of two additional minor fields. I next roll 53, so the sage's major field of study is humanoids and giantkind. I then roll 66 and 38, so the sage also has minor knowledge of physical universe(s) and demi-humankind. Tragically, there are no subtables to determine the special categories known to the sage, but there are eight listed under humanoids and giantkind, so I'll roll 3d8. I get 5 (legends and folklore), 4 (languages), and 3 (history).

Whether or not a sage knows the answer to some question is based on whether the information is within one of the sage's fields of study or special categories, as well as the nature of the question (general, specific, or exacting):

As you can probably imagine, determining whether a question is of a general, specific, or exacting nature is left to the judgement of the DM:

Here's where we get to the good stuff. In order to conduct their studies, sages require no fewer than four rooms (living quarters, a study, a library, and a workroom) and a total of at least 800 square feet of space for all their materials. The DM is specifically instructed to "make a point of asking for far more than is actually needed", with the example sage specialized in flora requesting an additional root cellar, greenhouse, fungi beds, and several acres for growing plants:

800 square feet is is eight 10' squares or 16 5' squares, so it's not a lot of real estate on a sheet of graph paper, but one can imagine a sage with even just the bare minimum amount of space required occupying at least an entire floor of a tower in, say, the player character's stronghold.

You can roll to determine the sage's ability scores, alignment, and hit points:

As you can imagine, sages have high Intelligence and Wisdom, average Dexterity, and slightly below average Constitution and Charisma. For some reason, they tend towards average Strength (11.5, the same as if you rolled 3d6), but with a higher minimum score (8). They also have 8d4 hit points before Constitution bonuses are applied, which is the same as an 8th level wizard! They are most commonly Lawful Neutral or Neutral in alignment, and most rarely Chaotic Evil or Chaotic Good.

Interestingly, sages are capable of casting spells, and which spells they can cast is determined by a die roll (d4+2 to determine the maximum level of spells, from 3rd to 6th) and their fields of study:

This also suggests that 70% of sages have clerical abilities, 20% have druidical talents, and 10% have magic-user or illusionist abilities (referring back to the table for determining fields of study).

My hypothetical sage with knowledge of humanoids and giantkind, physical universe(s), and demi-humankind is a natural fit for clerical spellcasting abilities. I roll 5 on d4+2, so they can cast up to 5th level spells and will know 1d4 spells of each level. Like a magic-user (or even a cleric, if you choose to take to heart a brief aside in OD&D), a sage's spells are "contained in various source books." This is thematically excellent and evocative, and is maybe the only (?) mention of druid spells being written down somewhere, which is neat.

I'm less sure what to make of the assertion that sages cannot cast "Spells such as bless, chant, prayer, commune, raise dead, commune with nature, and contact other plane." The prohibition against raise dead makes enough sense, as - if no one in the party is capable of casting it - the player characters are expected to pay out the ear to get an NPC to cast such a spell for them. The commune spells and contact other plane are thematically within the sage's wheelhouse, but their inclusion on the list makes some sense, as their use undermines the very purpose of the sage (and makes it far too easy for them to retrieve knowledge on the party's behalf). If other basic spells like bless, chant, and prayer are forbidden, it raises the question: What spells exactly are sages supposed to be allowed to learn? I wish the DMG offered some guidance on that front, but alas.

Note that, because the sage is the most annoying NPC in AD&D, they "will tend to keep [their] spell knowledge as highly secret." The player character is only providing employment and room and board, after all - why should the sage be upfront with them about their abilities?

The next paragraph is not especially remarkable, but take note that sages are all middle-aged to venerable in age:

We'll come back to that in a moment.

Only certain classes can hire sages:

The notable exclusions from the list are magic-users, illusionists, clerics, druids, and monks - which, aside from monks, are the very classes to which sages kind of sort of belong. Perhaps this indicates another aspect of the sage's particularness - they cannot bear to be in the employ of someone with abilities similar to or greater than their own.

Any character can "consult" with a sage (which will later be referred to as "short-term employment"), but any character of the aforementioned eligible classes seeking to employ a sage on a long term basis "must have a stronghold with ample space for the sage" (the four 200-square-foot rooms mentioned above, at the very least) and must be willing to employ the sage for the rest of the sage's life! Very demanding, these sages.

Sages are found only in large towns and cities:

How many sages might be found in such a place? We can work backwards to figure that out. If you have a means of identifying what character-type NPCs live in your settlements, you could decide that any middle-aged to venerable magic-user, illusionist, cleric, or druid in the area capable of casting 3rd to 6th level spells is available for employment as a sage. Using my B1 sandbox as an example, I might decide that Sigismund, a 7th level magic-user, is a sage.

You can then keep working backwards to determine such a sage's fields of study. Sigismund, as a magic-user, would presumably major in supernatural and unusual studies, with minor knowledge in other fields.

I'm fascinated by the assertion that sages belong to an informal "Brotherhood" (but note that they're not all assumed to be men), and that all sages in the area will know about the employment of a sage. The sages are hanging out at the forums, gossiping about all the dumb questions you ask them!

A player character who consults with a sage is employing them on a short-term basis, which is limited to one week's time. After which, the sage is unavailable for one month, because sages don't like to waste their time with "foolish questions" (especially if the fools aren't offering lifetime employment and room and board):

The one-week restriction on short-term employment will affect the sorts of questions a sage so employed will be able to answer, which we will get to later. Note that this service costs at least 100 gold pieces per day!

Sages employed on a permanent basis are also incredibly expensive, costing 400 to 2,400 gold pieces per month:

The "initial material expenditure" represents the purchasing of "reference works, experimental equipment, and so on." The minimum amount (20,000 gold pieces) allows the sage to answer specific and exacting questions at just 50% "efficiency" (unclear to me if this is referring to time spent researching the answer to the question or the sage's actual ability to find the answer). 60,000 gold pieces gets the sage to 90% efficiency, and only after 100,000 gold pieces are spent does the sage operate at 100% efficiency:

The player character can spend additional gold to increase the sage's question answering ability in the general and specific areas. The player character can spend up to 25,000 gold pieces to increase the sage's knowledge outside their fields of study by up to 5%, up to 50,000 gold piece to increase their knowledge in their minor fields of study by a like amount, up to 300,000 gold pieces to add three minor fields of study, and 200,000 gold pieces to add a major field of study! 

Payments are made up front. Increasing knowledge requires a month of game time per percentage point, and adding a field of study requires 2 years of game time. If the player character dares ask a question of the sage during this time, "all is lost" - the sage completely "forgets" all they have learned so far, and presumably demands more money and more time to begin their studies again.

Discovering information takes time, and this is determined by the type of question asked:

A sage can answer a general question in just a few minutes, but specific questions require hours, days, or even weeks. Exacting questions require days or weeks of time. Remember that a sage employed on a short-term basis charges a 100 gold piece fee per day in addition to the costs shown on this table, and that they will only accept employment for a week's time. Thus, the specificity of question which a sage so employed can answer is greatly limited.

There is a chance that answering the question will require more funds, if for example the sage must travel to some town or city for materials. There is also a small to good chance that the reference material is readily at hand, meaning that no additional expenses are incurred:

It is always best to find a sage who has knowledge of the relevant special category of study, but especially so if the sage is being consulted on a short-term basis.

One especially irritating characteristic of the sage is that they will "continue a fruitless search" for knowledge even if a question can never be answered positively:

This sounds as if the sage is not aware that the question is impossible to answer and is simply trying their hardest to find the information, but I choose to interpret it as the sage deliberately wasting the player character's time and money, because it fits my narrative that these NPCs are being annoying on purpose.

Researching a fruitless line of inquiry will always take the greater part of the maximum time required, but at least the sage will only charge half the usual amount for their ineffectual services:

I really like the image of the sage sitting and looking wise for 4 to 6 minutes before simply saying "Yeah I don't know."

Let's say the player characters seek out a sage to consult on a question of an exacting nature. The sage agrees to try to find the answer for them, but will only spend one week doing so. The subject is one of the sage's minor fields of study, so they demand 1,100 gold pieces per day to research it. I roll 5d8 and determine that it will take the sage 23 days to find the answer, so it is impossible for them to do so in the timeframe given. After a week, the sage meets with the party, shrugs their shoulders, and demands their payment of 3,850 gold pieces (half the agreed upon amount). Money well spent! And no, the sage will not try again or spend another week continuing their research. They have more important things to do. See you in one month!

This is my favorite part. After spending more than a day researching the answer to a question, the sage requires time off (1 day for every 3 spent working - quite the PTO policy):

If you bother the sage during their R&R time, they demand additional days off for the intrusion! Until you give them their time off, they will (presumably, deliberately) take the maximum amount of time and require the maximum amount of gold expenditure to answer any questions. The sage has excellent work-life balance, and the Brotherhood, while apparently informal, is presumably a strong and demanding union. Good for them!

Before you start to sympathize with the plight of the sage, let me add that they are also racist:

In a world of scammers and jerks, these guys really are the worst of the worst. You do not, under any circumstances, "gotta hand it to them"!

Using AD&D's Spying Table to Resolve NPC Missions

AD&D is a game in which the player characters are expected to overcome challenges themselves. This is made most explicit in the section on NON-PLAYER CHARACTERS, and is embodied by the guiding principle that dealing with NPCs "should be expensive and irritating" - that is, that player characters seeking aid from outside sources should be dissuaded from doing so by the mere fact that those DM-controlled characters will be extremely costly and annoying to recruit, convince, placate, and the like. This is made explicit in the section on HIRING NON-PLAYER CHARACTERS TO CAST SPELLS OR USE DEVICES:

I'm sure that there are many other examples in the text, although my Ctrl + F skills are failing me at the moment to dredge up more evidence. You will have to trust that I've read enough AD&D to be an authority on the matter when I say that this sentiment is apparent in Gygax's writing even when he doesn't spell it out in so many words.

Deserted castles dot the landscape, but do not exist solely for the convenience of player characters seeking strongholds of their own. There are roads to the dungeon built for ease of access, but they're not free to use. Towns and cities are full of NPC adventurers looking for work and high level NPCs with a great deal of power, but henchmen are not easy to find, and high level NPCs are more likely to use the player characters for their own ends than they are to aid them. Dragons' lairs contain mountains of treasure, but henchmen will not stay and guard them on the player characters' behalf. Player characters can find a cleric to cure them of their diseases, but it will cost them dearly.

One exception is, of course, hirelings - those NPCs which the player characters hire and pay on a regular basis (daily or monthly, in the case of standard hirelings, or monthly only in the case of expert hirelings). Most of these are not the adventuring sort, however, and mainly serve to do those mundane or tedious tasks on behalf of the player characters which would make for dull gameplay if handled by the player characters themselves. A limner can draw up a heraldic device for the fighter, an armorer or blacksmith can toil away in the player character's stronghold, and a properly equipped sage can conduct research to come up with the answer to some question in which the player characters have an interest.

Hirelings will go on adventures sometimes, but these are never without the party present. Bearers and porters might carry treasure from the dungeon back to town, linkboys might bear torches or lanterns, pack handlers and teamsters might deal with beasts of burden and vehicles, a scribe might keep the party's map of the dungeon, and a ship's crew might ferry them along a river or across a lake to expedite their forays to and from the site of adventure, but these are all supplemental to the party of player characters - never a replacement for them.

The one hireling which embarks on their own to engage in what we might recognize as "adventuring" activity is an exceptional sort of expert hireling - the Spy:

The spy is a type of assassin NPC which can be recruited much like any other hireling or henchman. They are similar to the sage in that they follow their own bespoke set of gameplay rules and in that their principle purpose is the obtaining of information. The difference is that - while yes, it would be cool if the information the party needs is located in an ancient tome inside a long lost library, and they must adventure to obtain it - the sage's purpose is to abstract the dull work of actually spending months of game time poring over the texts for the answer to the question, whereas traveling to a castle to gather information about the ruler's mustering of forces, infiltrating a secret society, or posing as important officials to gain access to their resources - all without being discovered - is a bona fide dangerous adventure that the player characters well and truly could engage in themselves, and it would be fun to do so.

Whether or not it's a good thing that this potentially fruitful adventuring activity can be handed off to an NPC is not really my concern. Why this particular adventuring activity can be handed off to an NPC and yet the party cannot, say, hire another party of adventurers to conduct some other mission on their behalf (or send their henchmen, their mercenary army, or other followers to do so) is a more interesting question.

It's not such a confounding question, really. The answer is that spying, for one reason or another, is an element of adventuring which Gygax felt comfortable allowing the player characters to engage in at a distance. Going into dungeons, clearing out monster lairs, and laying siege to strongholds is the meat of adventure, and player characters should be involved in those activities themselves.

Except they don't really need to - not 100% of the time. When a necromancer moves into some ruined tomb the player characters cleared out at 1st level and starts raising the dead, why shouldn't those same characters - now rulers of their domains with huge armies and companies of henchmen - be allowed to send those henchmen off on their own adventure to deal with the problem while they focus on more important matters? Why shouldn't they be able to delegate the work of clearing monster lairs from their territory to their mercenary army? When some petty lord in a humble tower stops paying taxes, why should we stop the game to go put him in his place?

The answer is, obviously, because that stuff could be fun. You can send your henchmen to the old dungeon to clear it out again - you just play as your henchmen. You can send the captain of your army to chase the goblins off your land or bring your vassals back into the fold - you just play as the captain commanding the troops. It's all adventure, and worth being a part of.

But maybe sometimes the players don't want to spend a bunch of game time playing out that sort of thing. Their characters have other stuff to do, and they want to get to it. Is that such a crime? Maybe. But I think it should be an option on the table, and if we can abstract spying missions, surely we can abstract other adventuring activity involving NPCs, and I think AD&D's ASSASSIN SPYING TABLE provides a pretty good framework for doing so:

I would argue in favor of ignoring the specifics of this table (that is, the fact that it pertains to spying missions in particular) and instead focusing on what it represents more broadly. On the Y axis, we have the level or strength of assassin, spy, or, more generally, the NPC(s) undertaking the mission. On the X axis, we have the degree of difficulty of the mission. The Simple, Difficult, and Extraordinary categories are defined as such:

While these are all defined within the framework of spying (what sort of action is required of the spy and the resulting chance of discovery), I don't see why they can't be generalized.

For example, the Level Of The Assassin/Spy might instead be the average level of the henchmen sent to clear out the dungeon, or the level of the commanding officer leading the army to clear out the monster lair or lay siege to the stronghold. 

If henchmen are sent to clear out a dungeon, whether the mission is Simple, Difficult, or Extraordinary could be determined by the level of the dungeon relative to the level of the henchmen. If the henchmen are, on average, 4th to 7th level, a dungeon of 3rd level or lower might be a Simple mission (65 to 80% chance of success), a dungeon of 4th to 7th level Difficult (40 to 55% chance of success), and a dungeon of 8th level or higher Extraordinary (20 to 30% chance of success).

If an army is sent to clear out a monster lair or lay siege to a stronghold, an owlbear den or small stronghold (a small shell keep, tower, moat house, or friary) might be Simple, an ogre lair or medium stronghold (a large shell keep or small or medium walled castle) might be Difficult, and a village of 300 orcs or a large stronghold (a concentric castle, large walled castle, or fortress complex) might be Extraordinary. Such an army led by a 5th level captain would have a 25 to 70% chance of success, while the same army led by an 8th level captain would have a 35 to 85% chance of success. You can also use your understanding of the complexity of different location types when determining a mission's difficulty.

We're also given the amount of time required to accomplish the mission:

This does not include travel time to and from the mission location. Simple missions require 1d8 days (average 4 or 5 days), Difficult missions require 5d8 days (average 22 or 23 days), and Extraordinary missions are left to the referee's discretion. If we want to proceduralize it, we could call it 9d8 days (9 to 56, average 40) or 10d8 days (10 to 80, average 45 days). That seems fair. If the players want it done faster, they should do it themselves. This also ensures that they are limited in the number of missions they can have their henchmen, hirelings, and followers conducting at any given time. If players rely too much on these resources, they will soon find themselves spread thin.

In spying missions, there is always a chance that the spy is discovered:

The chance of discovery is a function of the time spent spying (longer missions are more dangerous), the level of the spy (the use of low level spies is more dangerous), and the precautions taken by the spied upon (more prepared targets for spying are more dangerous). Being discovered automatically results in mission failure and imposes a modifier to the SPY FAILURE TABLE which generates more disadvantageous results on that table.

Since this is very specific to spying, and the missions which I'm talking about are not concerned with being "discovered", we might recontextualize this mechanic instead as a "chance of complication". The mechanic is already very much left up to the DM's judgment, so we can get weird with it, using these modifiers to account for any other mitigating factors not already accounted for by our initial calculation of the mission's chance of success - that is, any other considerations which might negatively impact the mission's chance of success.

In the case of the dungeon, are there undead present, and does the dispatched party lack a cleric? Are there monsters which can only be harmed by magic weapons, and does the dispatched party lack these tools? Does the dispatched party have a map of the dungeon? Have wards and traps beyond the ordinary been placed there? Is there some higher level monster present of which the party is unaware?

If the target is a monster lair, are the monsters organized by some more intelligent force than is known? Are the monsters expecting an attack? Do they possess some especially powerful magic item? Have they made an alliance with other monsters in the area? Is the terrain especially unforgiving towards a large army? 

If the target is instead a stronghold, is it ruled by a powerful spellcaster? Does the ruler keep a pet dragon? Does the garrison possess other monstrous creatures like griffons or ogres (such as those strongholds detailed in OD&D)? Is the stronghold well provisioned to withstand a siege? Does the ruler have a spy of their own among the player characters' followers? Is an allied army rushing to meet the besiegers in the field?

The last bit to consider is the SPY FAILURE TABLE and its modifiers:

Again, we can loosely interpret these results to suit our purposes:

• On a result of 01 to 35, the mission simply fails and can be reattempted, but the duration must be rolled again. The time spent on the first attempt is wasted. 
• On a result of 36 to 60, the mission fails, and further attempts are 90% likely to fail as well. We can interpret this as the denizens of the dungeon, monster lair, or stronghold taking additional measures to thwart further missions against them. 
• On a result of 61 to 80, the player characters' agents are caught, imprisoned, and not heard from again.*
• On a result of 81 to 95, the agents are caught and tortured. In the process, either they are killed, they reveal some information which the party does not wish to fall into enemy hands (something about the party's plans or interests, or maybe a vulnerability of their own stronghold), or they are turned to the enemy's side.*
• On a result of 96 to 100, the agents are simply killed or turned to the enemy's side.

*A result of 61 to 95 can also result in the agents being set free, but only if doing so would disadvantage the party in some way. Perhaps they are cursed or infected with a disease and sent back to the player character's stronghold to spread it among their other followers, or perhaps they are unknowingly given a device by which the enemy can scry upon the characters and monitor their activities.

As you can see, "discovery" (or in our case, an unforeseen complication) greatly increases the chance of something bad happening as a result of mission failure. Difficult missions are also more likely to yield unfortunate results when failed. Missions of Extraordinary difficulty are for some reason less likely to result in a bad outcome upon failure (still trying to wrap my head around that one).

It is also worth shouting out the brief aside on Fanatical Spies:

This is very flavorful and very spy specific, but we can use it. Rather than killing themselves (which is the bit that feels very specific to spies), perhaps certain NPC followers are so loyal and devoted to their player character masters/causes that they simply refuse to allow themselves to be captured or turned, forcing their enemies to torture or fight them to the death, rather than betray the player character. This would be an excellent opportunity for a loyalty check.

This bit also points us back to the EXPERT HIRELINGS section (at the top of this post) to determine how much it costs to employ a spy. While we're concerned with henchmen and hirelings already in the player characters' employ with some sort of set compensation (either shares of treasure or a certain amount of coin per month), it does make sense that undertaking an especially dangerous or involved mission such as this would warrant some kind of bonus pay.

Instead of referencing the information about spies, however, let's flip over to the PHB and the table MINIMUM FEES FOR ASSASSINATION:

This table lists the amount of gold earned by a player character assassin for a successful assassination. This is a function of the assassin's level and the level (or notoriety) of the victim. I imagine it could also be the amount of gold the player characters must pay to hire an NPC assassin to do their bidding - after all, why not? Might as well make use of those assassins' guilds running around AD&D towns and cities.

We can also use this as a reference for roughly calculating what sort of "bonus pay" player characters might have to pay out to their followers for completing some mission. The 3rd level henchmen sent to clear out a like-level dungeon would receive a bonus of 225 gold pieces (plus any plunder from the dungeon itself, presumably), the 5th level mercenary captain sent to lay siege to a 9th level fighter's stronghold would receive 900 gold pieces, and the 8th level captain sent to battle an orc army would earn somewhere around 4,000 gold pieces to distribute amongst the soldiers under his command.

Getting others to do your dirty work isn't always fun, but at higher levels of play, player characters have access to lots of resources, and lots of options when choosing what they'd like to do next. I don't see anything wrong with letting them delegate some of their responsibilities so that they can prioritize the goals in which they're personally invested without the world falling to ruin around them. I for one think it's neat that player characters with enough resources can simply hire a spy to infiltrate a secret society on their behalf, or pay an assassin to go kill that evil wizard for them so that they can do something else. 

I don't see a system like this getting much actual use at the table. Most campaigns probably won't reach this level of play, and when they do, I imagine that most players will want to play out these scenarios themselves (whether as their own characters or by assuming the roles of their followers), but I think it's interesting food for thought and wanted to demonstrate how it might work in practice.

Of course, you don't have to limit yourself to resolving player character-orchestrated NPC missions this way. You can also use this method to resolve other NPC "missions", sort of like a "faction turn". Does the rival adventuring party clear out the dungeon and steal the player characters' glory out from under them? Does the villainous wizard successfully create that new magic item or spell they've been working on? Does the upstart gang of thieves in town overthrow the reigning Guild? There are many potential uses. 

AD&D is full of these odd little resources that you don't see very often in modern TTRPG materials, usually covering some very specific scenario which doesn't come up all that much in play. It's fun to let these odds and ends inspire you and let your imagination run wild with creative new ways to utilize them. 

Good Henchmen are Hard to Find

Apparently my special interest now is writing about the annoying mundanities of AD&D city/town life - taxes, tolls, intentionally misleading encounters, and irritating and expensive NPCs. To this end, I've been reading all about henchmen in the DMG. Boy, it is a lot more difficult to recruit these people than I gave it credit for!

Henchmen are not to be confused with standard hirelings like porters, teamsters, and torchbearers, nor with expert hirelings like mercenaries. Henchmen are character-type NPCs - that is, those with at least (and often only, at least when initially recruited - more on that later) one level in a class who directly strengthen the party by (mostly) loyally serving under the direct supervision of a specific player character.

Not only do they serve as another character-type "body" in the party. They can also strengthen the player character's stronghold, serve as a backup character to adventure when the player character is unable to (including when the player character is dead), and, perhaps most interestingly, be used "as a safety measure against the machinations of rival player characters":

It's no surprise that in modern editions of D&D - where player characters are independently strong, strongholds are non-existent or capriciously handwaved, character death is rare and easily recovered from, and player characters are generally assumed to be working together amicably - henchmen are more or less a thing of the past. But in AD&D, henchmen are explicitly "so useful" and "so devoted" that a player character's Charisma score imposes a limit on the number they can have in service at any given time:

This assertion that henchmen are incredibly useful is important to keep in mind, as it provides justification for just how hard it is to recruit them.

I'll provide a bit of perspective as to where it is I'm coming from. Typically in my games, when the player characters are looking for henchmen, they go to a nearby settlement, spend a week of downtime searching for candidates, I tell them who is available, and they pick who they want. I have a list of available henchmen in each settlement. Simple as that. I never considered that finding and recruiting henchman could be a challenge in and of itself, involving the expenditure of time and money and the use of social acumen. 

Whether it should be such a challenge will depend on the individual DM's taste. Since I am recently enamored with Gygax's assertion that dealing with NPCs "should be expensive and irritating", I am currently of the opinion that jumping through hoops to recruit henchmen is similarly a somewhat charming artifact of Gygaxian design sensibilities. That being said, let's get into the nitty gritty.

To begin with, newly recruited henchmen are typically 1st-level:

Player characters above 6th-level might get a 2nd-level recruit, and characters above 11th-level might get a 3rd-level recruit at best. The parenthetical justification suggests that while player characters seek out henchmen for all of the reasons I've already mentioned, henchmen seek out player characters for protection ("the aegis of a PC") - higher-level character-types need no such protection and so do not seek out such work.

What follows are a few paragraphs about the race of potential henchmen:

The race of the player character somehow affects which henchmen will seek them out. This isn't expanded upon here, but I imagine that the DM is meant to reference the racism table RACIAL PREFERENCES TABLE in the PHB:

For example, dwarf player characters probably won't attract elf or half-orc henchmen and vice versa.

Racial distribution of the recruitment area is up to the DM's discretion. I personally use this bit from APPENDIX C:

I assume that my settlements are largely human. Dwarves, elves, gnomes, and halflings don't have their own "settlements" per se, they have "lairs" (small-scale ethnocentric communities) as determined by stocking the campaign area using the wilderness encounter tables (for example, a few demihuman/humanoid lairs pop up in the second part of my series on using the DMG to stock a sandbox). You can assign your own bespoke racial distributions to each of your settlements as you like, but that sounds to me like more trouble than its worth, so I use this blanket distribution.

Yes, you can racially profile when recruiting henchmen. You can put up posters all over the city that say "HENCHMEN WANTED: Half-Orcs Need Not Apply!":

Very cool!

Now that we've covered the fun issue of race, let's find out how many henchmen are in a given settlement:

In my post on settlements, and again in the first part of my series on stocking a sandbox with the AD&D DMG, without ever reading this section, I suggested that probably 1 in 100 people in a settlement would be character-type NPCs. I was pretty close! I just didn't account for the possibility that non-humans (and non-half-orcs) would be twice as likely to be character-types. Why half-elves are twice as likely as half-orcs to be "suitable for level advancement" is beyond me (it's racism again, I guess).

Of those 1 in 100 humans and half-orcs, one-tenth are recruitable as henchmen (the rest are either higher-level or "already in a situation they are satisfied with" - perhaps already in another character-type's employ or otherwise adventuring on their own). I'm going to assume then that one-tenth of non-human non-half-orcs are also recruitable (so 1 in 500 of the non-human non-half-orc population, compared to 1 in 1,000 of the human and half-orc population).

As an example, let's crunch the numbers for the average city. I'll refer to APPENDIX B's INHABITATION table:

Typically we would roll 1d6 and multiply the result by 10,000. The average of 1d6 is 3.5, so let's say our population is 35,000 (even though you can't actually roll 3.5 on the die, this is a hypothetical exercise). Our racial distribution is:

• 28,000 humans (80%)
• 1,750 dwarves (5%)
• 1,750 elves (5%)
• 700 gnomes (2%)
• 1,750 half-elves (5%)
• 700 halflings (2%)
• 350 half-orcs (1%)

Here's how many character-types there will be:

• 280 humans
• 35 dwarves
• 35 elves
• 14 gnomes
• 35 half-elves
• 14 halflings
• 3 or 4 half-orcs (50/50 chance)

Here's how many recruitable henchmen there will be:

• 28 humans
• 3 or 4 dwarves (50/50 chance)
• 3 or 4 elves (50/50 chance)
• 1 or 2 gnomes (60/40 chance)
• 3 or 4 half-elves (50/50 chance)
• 1 or 2 halflings (60/40 chance)
• At most 1 half-orc (30 or 40% chance)

Since the highest-level a newly-recruited henchmen can be is 3, I would roll d3 to assign a level to each or else divide by 3 and roll for the remainder. For example:

• 15 1st-level (9 humans, 1 dwarf, 1 elf, 1 gnome, 2 half-elves, 1 halfling)
• 14 2nd-level (10 humans, 1 dwarf, 1 elf, 1 half-elf, 1 halfling)
• 14 3rd-level (9 humans, 2 dwarves, 2 elves, 1 half-elf)

This is actually kind of useful. Until the player characters attain 7th-level, in the average city you need only prep 15 character-type NPCs to be potential henchmen. At 7th-level they might attract a 2nd-level henchmen, but even then it's only a 10% chance, so you could prep just one for every nine 1st-levels. At 12th-level you could prep one 2nd-level and one 3rd-level for every two 1st-levels (since the distribution is 25/25/50, respectively). 

Furthermore, if only 1 in 500 to 1 in 1,000 people are recruitable as henchmen (and the 1 in 500 applies only to 20% of the population), the smallest settlement which will definitely be home to recruitable henchmen is a town (the average of which will only have 3 to 5). In smaller settlements, you could determine the number of character types and then dice to determine if any henchmen are available. For example, the average village will have 750 people - that's 6 human or half-orc character-types and 1 or 2 others, so a 60% and 10 or 20% chance of a recruitable henchmen, respectively.

You still might want to find out who the other character-types in the settlement are (I recommend fleshing out the most powerful people in the settlement - those of the highest level in a small settlement and any who might rule strongholds in larger settlements), but this has already greatly reduced the amount of character-types we need to prep in our settlements.

So, math-y diatribe aside, how do you actually recruit these people? Surely you just go to a tavern and they're all hanging out there, waiting for you, right? ...Right?

Wrong! You need to do more math:

If you somehow fail to recruit henchmen (or fail to find the henchmen you want), you have to wait a month before trying the same method again. You can combine methods for a better chance at recruiting, but it has diminishing returns. All of this is expensive. In the example provided, the player character spends 860 gold and attracts 65% of available henchmen.

Let's assume that this is done in our example "average city" with 43 recruitable henchmen levels 1 to 3. If the player character is below 7th-level, let's say they attract 65% of the 15 1st-level henchmen (9 or 10 recruits). If the character is 7th- to 11th-level, they can attract 65% of the 15 1st-level and 14 2nd-level henchmen (18 or 19 recruits, with 1 or 2 - 10% - of those being of 2nd-level). If the character is above 11th-level, they can attract 65% of all 43 available henchmen (27 or 28 recruits, with 6 to 7 each - 25% - of those being of 2nd- and 3rd-level).

Let's say 9 1st-level recruits are attracted by a relatively low-level player character. Using our example demographics from above, let's say 7 are human, 1 is a half-elf, and 1 is a halfling.

Of course, you have to wait around for a few days for all the recruits to show up. If you're a no-show, they leave:

I determined it will take six days for all recruits in our example to show up - one on each day plus three on the second day. That's almost a week where the player character can't get much of anything done because they're just waiting around.

Whatever they do, our example player character had better not ensorcell the recruits or ask them about topics which are not polite conversation (alignment and religion):

I hope our example player character isn't expecting anything fancy like a monk, assassin, druid, or illusionist:

The percentages break down like this:

• 16.67% cleric
• 3.33% druid
• 35.2% fighter
• 4.4% ranger
• 4.4% paladin
• 16.67% magic-user
• 3.33% illusionist
• 12.5% thief
• 2.5% assassin
• 1% monk

Which is only slightly different than the Character Subtable used for dungeon encounters in APPENDIX C:

Clerics, fighters, and magic-users are slightly less common, but everyone else is more common. I appreciate that the two tables are pretty consistent, although I don't really see the need for two in the first place.

I feel similarly about the note on multiclassed henchmen:

The section on Race and Multi-Class in APPENDIX C (scroll up a bit for the screenshot, since I used it earlier) provides a totally different method for determining if a character-type NPC has two or even three professions. Why are there two slightly different methods for generating character-type NPCs on the fly - one for those in dungeons and one for henchmen recruits? I do not know.

Here are the nine recruits who shows up while our example player character is waiting around:

• Day 1: LE human assassin
• Day 2: N human druid, CG human fighter, CE human fighter, CN human illusionist
• Day 3: NE human magic-user
• Day 4: CG half-elf cleric/magic-user
• Day 5: LG halfling fighter
• Day 6: CE human fighter

Our example player character got lucky, managing to attract an assassin, a druid, and and illusionist in addition to the more common types.

Let's say that the player character has advertised, the recruits have arrived, and it's time to make an offer:

You have to offer at least 100 gold per level of the henchman (300 gold max for a 3rd-level henchman) for them to even consider the offer, and that only gets you a 25% chance that the henchman will accept! You can get that all the way up to 55% for 300 gold more. That's right - the best you can do is offer a 400 to 600 gold signing bonus, and the henchman basically flips a coin to decide if they accept or not!

You can increase the henchman's interest by offering magic items:

You also have to provide all of their equipment! Why don't they have equipment? Well, they're wretches. That's part of the reason why they're seeking employment:

Pity the wretched henchmen of the AD&D world!

This section seems to suggest that 3rd-level henchmen (not 2nd-level henchmen) will come with equipment:

This also suggests that player characters can have henchmen (or rather, "associates") of a level higher than 3rd (and even of a level higher than the player characters themselves), but only temporarily in most cases. This seems to be describing the acquisition of henchmen/associates by means other than the standard recruitment process, as the NPC will become a regular (non-temporary) henchmen only if they are three or more levels under the level of the player character, which is a requirement unique to this situation.

Interestingly, this means that, for example, an 11th-level player character could potentially acquire an 8th-level henchmen through this route. I imagine this "exceptional henchmen" rule is used if, for example, the player characters topple a bandit stronghold and wish to recruit the captured character-type bandits to their cause.

Be sure to offer your henchmen housing, food, and clothing! Doing so only adds 5% to their interest level, but if you neglect to offer such support (or simply don't know that you're expected to), the recruit's interest level drops by 25%:

With the bare necessities like the signing bonus, equipment, and other support out of the way, the recruit will want to know what exactly their duties will entail and what their share of treasure will be:

Here I wish there was a little more detail. Is it not assumed that the henchman's position will be subordinate to the player character and that their duties will involve going into dungeons and braving traps and monsters? Is the henchman more or less likely to accept if they're given command of other henchmen? What about if they will be garrisoning a stronghold rather than adventuring? What is the expected share of treasure? What if less or more is offered? With such thorough percentage breakdowns elsewhere in this section, why not here as well?

I have similar thoughts about the provision of equipment. "The prospective henchman must be provided with complete equipment according to his or her class or classes." Will a fighter simply accept leather armor? Will they expect at least chain mail? Will plate armor sweeten the deal? It's easy enough for the DM to simply use their own discretion here, but it is odd that the guidance is less concrete in some areas than others.

The DM should also use the section on NPCs to determine henchmen characteristics. Hopefully the player characters don't try too hard to get to know their henchmen before they're hired, as this would be considered a faux pas:

Whatever the henchman's characteristics and alignment, remember that dealing with them should always be expensive and irritating!

Let's say our example player character is a CG human fighter. They meet with the LE assassin on day one and it's clear that the two don't get along. They wait an additional day and meet with the four recruits who arrive on day two. They get along best with the other CG human fighter, but the druid and illusionist fill roles that aren't already represented in the party at large (and they're harder to find), so the fighter wishes to recruit them.

Finally, with all else being known, the DM tallies up the henchman's interest level, adds the player character's Charisma reaction adjustment, and rolls to determine if the henchman accepts employment:

Let's say money isn't an issue, so the fighter offers the druid and illusionist each 400 gold as a signing bonus. They don't have any magic items to offer, but they're wise enough to offer to pay for food, clothing, and lodgings, making the chance of acceptance 60%. Their Charisma score is 13, which adds 5% for a total of 65%. For whatever reason, the druid balks (I rolled an 81 on d100, so perhaps negotiations over a couple of magic items was the point of contention), but the illusionist accepts. The fighter could stick around to see if more recruits come in, but they're satisfied with the hire and want to get on with adventuring.

But that's not all! Once a henchman is successfully recruited, the player character must keep the henchman. This is where loyalty comes into play:

The henchman's loyalty is tested in certain situations. If they fail their loyalty test, the henchman might co-operate with or surrender to an enemy, testify against their liege (suggesting that yes, the DM should put their player characters on trial for their crimes), steal, desert, refuse orders, or run away:

The loyalty score is 50% plus or minus the player character's Charisma adjustment:

For context, here is the CHARISMA TABLE from the PHB:

The reaction adjustment is used when recruiting henchmen, but the loyalty base is used when determining, well, base loyalty of henchmen. With the fighter's 13 Charisma, the newly-recruited illusionist's loyalty base is a flat 50%.

There are tons of modifiers to a henchman/hireling's loyalty base. Henchman are actually the most loyal of all NPC followers, and even they only have +5% loyalty:

However, henchmen are also treated as having the highest level of training (comparable to officers or major officials), granting another +30%:

The illusionist's loyalty to the fighter is now up to 85%!

The longer a player character knows a henchman (and thus keeps them comfortable, fed, equipped, and of course, alive), the greater their loyalty:

This brings the illusionist's loyalty down to 80%, since they've only just now been recruited.

This is also where shares of treasure come into play:

I assume that an equal share of treasure is "average". Would it be fair for shares of treasure to be weighted based on relative level? That is, would it be acceptable for the 1st-level henchman of a 4th-level player character to receive a 20% share of the character's treasure? I'm not sure. Whatever it is, player characters must also pay their henchmen 100 gold pieces per level per month, as described under PLAYER CHARACTER EXPENSES:

Let's say the fighter is 1st-level and offers the illusionist an equal share of treasure. Loyalty is unchanged.

As the two adventure together, discipline, activity, and general treatment will come into play:

Being cruel and domineering towards your henchmen is actually a bonus - so long as you are present, alive, and conscious!

As the fighter is CG, let's say discipline is lax, but treatment is just, kind, and invariable. The illusionist's loyalty is now 90%.

Racism, of course, must be a factor:

I've determined that the fighter is in a party with three other humans, who are all "preferred" by the human illusionist. I'm not sure how to interpret the bit about adjustments being cumulative. Since both the fighter and their allies are all human, do I add 20% or 35%? It seems a bit much to count both hefty bonuses, so let's just count the 20%. The illusionist's loyalty is up to 110%.

Next we consider alignment:

The fighter is CG, which is -5%, but only one place removed from the illusionist (CN), which has no effect. The fighter's allies are CN, N, and N, which likewise has no effect. The illusionist's loyalty is at 105%. That's really good! It means the illusionist will not betray the fighter...unless the fighter breaks oaths, is reputed to have tortured, killed, or left henchmen to die, or actually tortures or kills them:

This is really interesting to me because it provides a means by which adversarial NPCs can spread rumors about the player characters which actually affect their henchmen's loyalty. On the flip side, the fighter can further improve the illusionist's loyalty by giving them gifts, rescuing them, or raising them from the dead. "Yeah, I know there's that insidious rumor going around that I tortured and killed my last henchman. Here, have some healing potions to make up for it."

Of course, there are situational modifiers to be used in combat as well:

Let's say the party encounters eight orcs on the first level of the dungeon. The fighter has been incapacitated due to damage but is still alive (hors de combat, -15%). The party has four 1st-level henchmen besides the illusionist, and two are dead or incapacitated (-10%). Additionally, one of the 1st-level player character is dead (-9%, counting the two dead henchmen). Six of the orcs are alive (-6%) and two are dead (+2%). The two remaining henchmen are alive and still engaged in combat (+10%). While the fighter is still technically alive and in sight, I'm not sure the 15% bonus would apply. That modifies the illusionist's loyalty by -28%, setting it at 77%. On d100, they roll 27, which keeps them in combat for now.

It wasn't easy, but it seems our fighter found a good henchman after all!

Hopefully in this post I have illustrated a few things. First, good henchmen are hard to find. It is both expensive and time consuming to do so, and they are in relatively short supply in all but the largest settlements. There is no telling who will show up and whether they will accept the player character's offer of employment. Because it is considered bad form to try too hard to get to know your henchmen before you hire them, the actual quality and compatibility of the henchmen is difficult to discern.

Second, good henchmen are hard to keep. Not only do they have to be paid well according to their level, but they must also be given their fare share of treasure, disciplined firmly but fairly, and treated justly, kindly, and invariably. You can improve your henchmen's loyalty by keeping them alive (and raising them from the dead) and by gifting them magic items, or damage it severely by breaking oaths or, well, torturing and killing them. Even then, there are factors that are outside of your control, such as personality, alignment, and uh...what race you are. Your enemies can spread rumors about you which will sully your reputation in your henchmen's eyes.

Of course, even if you do everything right, there is still a chance that your henchmen's loyalty will falter in a life-threatening situation. When your allies are dead and you're surrounded by enemies, even the most loyal henchmen might abandon you to your fate.

All that being considered, is it worth it? Is it fair to describe henchmen as "so useful" and "so devoted" as to require all this effortful recruitment and maintenance? That is for you to decide. Henchmen are a force multiplier. They are, in essence, akin to an additional player character in the party, and player characters are powerful. They become more powerful with the accumulation of levels and magic items. They can command a stronghold in your stead, fill in on adventures when your character is unavailable (or even replace the character if they die), and protect you against the "machinations" of other player characters (who are, of course, your rivals).

While I've demonstrated that henchmen can be disloyal based on a variety of factors, it is important to note that of all the followers you might recruit, henchmen's +35% loyalty base is the highest. While they're loyalty will vary in other ways depending on circumstances, they are otherwise "so devoted" in comparison to the other options available.

Perhaps this loyalty is born out of dependence. Henchmen are, after all, the wretches of the adventuring profession. While 90% of character-types found in settlements are of a higher level or "already in a situation they are satisfied with", henchmen are the bottom 10% who are destitute to the extent that they have no equipment and must seek employment, food, lodging, and protection from those more successful than themselves. It would make sense not to trust such individuals fully, but it also makes sense that they would recognize, when paid and treated well, that they have a good thing going and that they should cling to it with all their might.