In GM Don’t List #9: Fudging I discuss why GMs should avoid fudging, and if they do end up needing to fudge, why they should view that as a failure point in their game and a learning opportunity to figure out how they can become a better GM.

The most common form of fudging is changing the outcome of a die roll (the roll was a failure, but you say it was a success, or vice versa), but it’s widely understood that there are also other ways to fudge mechanical results (increasing a creature’s hit point total, for example).

What I think is less commonly understood, however, is that not all the dice rolls you make in a roleplaying game are mechanics, and it’s not actually fudging to change or ignore those dice rolls. Specifically, procedural content generators. Such generators can actually use any number of randomization techniques (for example, here’s a method for using CCG cards to generate adventures), but since we’ve already got dice laying around the typical RPG table most procedural content generators just use those.

WHY ISN’T THAT FUDGING?

If you’re struggling to understand why changing the outcomes of a procedural content generator isn’t the same thing as fudging a mechanical resolution, let’s take an extreme example. I’m prepping a scenario for my next session and I need a name for an NPC. So I pop open the Random Name Generator at Behind the Name, select for random surnames, click the button, and get:

Ivonne Eógan Masson

For whatever reason (maybe personal aesthetic, maybe because the Masons are already established as major power brokers in the city and I think it’s interesting this random generator has unexpectedly connected this NPC to the clan), I decide to drop the second “s” from “Masson” and name the character Ivonne Eógan Mason.

Did I just fudge?

Frankly speaking, no. Not by any reasonable/functional definition of the term.

What if instead of tweaking the outcome I actually just ignored it and rolled again by hitting the “Generate a Name!” button a second time? Still no.

What if I move this interaction from prep to actual play (I need to come up with a new NPC’s name on the fly, so I randomly generate one and then tweak it)? Still no.

What if the random name generator is published in the game’s rulebook? Still no.

This isn’t fudging not only because it would make the concept of “fudging” so broad as to be meaningless, but also because treating the outcome of a procedural content generator as a straitjacket or legally binding contract is to fundamentally misuse the procedural content generator. Using a procedural content generator is more like coating the bottom of an agar plate with a growth medium: As it’s exposed to your creative subconscious, the growth plate begins to accumulate a bunch of random creativity and odd synchronicities that begin to grow and thrive. (Ivonne Mason, for example, is a very different character than Lea Colton or Caroline Bone specifically because each of those random names provides a different creative stimulus.) Treating the outcome of the procedural content generator as if it were inviolable scripture, on the other hand, is like sterilizing the agar plate; it completely short-circuits the process.

ALL MECHANICS ARE PROCEDURAL CONTENT GENERATORS!

“Ah ha!” you say. “But aren’t all resolution mechanics actually procedural content generators, the results of which are meant to be creatively interpreted by the GM and other players? Is not the narration of outcome the same thing as taking a randomly generated group of bandits and creating the Blood Shield Bandits?”

Basically, no. There’s a similarity of process (roll dice, interpret results), but the function of resolution mechanics and procedural content generators are in many ways actually inverted: A resolution mechanic takes generally non-specific creative input and creates specificity (often literally a binary pass/fail state). A procedural content generator, on the other hand, produces non-specific creative input and expects the GM to create the specificity.

Because the processes involved are similar and because “specificity” can be a sliding scale, you can use procedural content generators as resolution mechanics (sterilizing your agar plates) and vice versa. But because the tools are designed for one thing and you’re using them for something else, the result is usually like using a screwdriver as a hammer.

GRAY AREAS

This is not, however, to say that there are no gray areas which lie along the boundary between resolution mechanics and procedural content generators.

You can see this perhaps most clearly when a game takes something which is traditionally a procedural content generator in other systems and makes it a hard-coded mechanic (or closer to being a mechanic) instead.

For example, in Apocalypse World characters are created from playbooks. For example, if you want to play a Gunlugger, you take the Gunlugger playbook and it instructs you, “To create your gunlugger, choose name, look, stats, moves, gear, and Hx.” Each of those categories then has a specific list of things. This becomes a surprising gray area: Many people, conditioned by other RPG character creation systems, looked at the provided list of names as a resource that could be used or ignored. (Many editions of D&D have similar lists of elven names, for example. Over the Edge provides a list of Al Amarjan names. And so forth.) Apocalypse World, however, specifically seeks to enforce setting through non-traditional mechanics, and so I’ve played at tables that instead interpreted this as a mechanical requirement: You must choose your name from the provided list of Gunlugger names (which are distinct from the list of, for example, Hardholder names, thus asserting setting). I’m actually still unsure what D. Vincent Baker’s intention was.

You can find another gray area in the dungeon stocking procedures of the original edition of Dungeons & Dragons. While the rules allow for some discretion on the part of the GM in the distribution of treasure and monsters, the line between advice, procedural content generator, and actual mechanic is very fuzzy and open to a lot of interpretation. This is even more true when it comes to wandering monsters in OD&D: A random encounter check is often interpreted as a procedural content generator by modern GMs, but in OD&D it’s stated as a straight-up mechanic.

The Mythic Game Master Emulator is also an interesting example: In order to emulate the role of the GM, the Mythic system basically adds a lot of binding structure onto a suite of procedural content generators. But despite the gray area this creates, the system still draws a fairly strong distinction between the output of the resolution mechanics and the output of the GM emulator (i.e., the procedural content generators). If you want to get a really clear feeling for how using resolution mechanics and using procedural content generators differ from each other, spending an afternoon playing around with Mythic as a cap-system for your favorite RPG can be very illuminating.

Go to Part 1

Using smart prep to improve your game isn’t just about the big, flashy stuff. In fact, the big, flashy stuff is often the easiest stuff to improvise while running the game. It’s actually the small, subtle stuff that requires thought and precision to implement effectively but also adds great depth to the players’ experience that can benefit the most from the care and consideration of smart prep.

One example of this is what I refer to as the exposition drip: When you have a significant bit of lore that you want the PCs to learn, but rather than delivering it all at once, you break it up and deliver it in chunks over time. This is can happen over the course of a single scenario, but you can also pace it out over the course of an entire campaign.

There are a few reasons for doing this:

• It allows you to pace the revelation to match the procedural plot. (This is particularly useful for mysteries, where the players consistently feel as if they’re making progress as they collect the puzzle pieces that add up to the ultimate conclusion.)
• By breaking the information up into smaller pieces, delivering them over time, and, thus, consistently revisiting the same bit of lore (without it becoming repetitive), you make it more likely that the players will learn and retain the lore.
• When you want a particular moment to pack a punch (“You have just discovered the lost sword Excalibur!”) it’s best if the players already understand the significance of what’s happening (what Excalibur is, who King Arthur was, etc.) before the punch is thrown. Preferably long before.
• On a similar note, breaking the lore into multiple parts allows it to be given to the players through multiple delivery mechanisms. This can make it substantially easier to make the players actually care about the information. (See Getting the Players to Care for a much longer discussion of this.)

And the reason this benefits from prep is because figuring out how to structure the exposition and the pace at which it should drip out benefits from pre-analysis and specific planning: If you mis-structure the exposition breaks, then the players can get ahead of the drip. If you pace the drip wrong, the players can lose interest or focus on the topic.

CREATING THE EXPOSITION DRIP

The process for creating an exposition drip is fairly straightforward.

First, identify the information they need.

Second, break that information into discrete pieces, each of which can be defined as a single revelation.

Third, determine how the PCs will learn each revelation. Ideally, set things up so that they can gain the information actively instead of passively (show instead of tell).

The way in which you actually do this is ultimately more art than science, though. For any given exposition drip, it’s a complex alchemy of the information being conveyed and the circumstances of the scenario or campaign in general.

NO, I AM YOUR FATHER

He’s a faux-media example of an exposition drip. You want one of the PCs to discover that their father is actually Darth Vader. This becomes a sequence of revelations, delivered as an exposition drip:

• Darth Vader is the Emperor’s right hand. He’s a really bad guy.
• Luke’s father, Anakin Skywalker, served with Ben Kenobi and was a hero during the Clone Wars.
• Anakin Skywalker was turned to the dark side of the Force and became Darth Vader.
• There’s still good in Darth Vader. It’s possible that he could be turned back to the light side of the Force.

Let’s consider, for a moment, the alternative to doing an exposition drip for this information. Luke follows R2-D2 to Ben Kenobi in the deserts of Tatooine.

Luke: You fought with my father in the Clone Wars?

Kenobi: You may have heard of Darth Vader. No? He’s a powerful Sith lord who serves the Emperor, and he’s also your father. He was my apprentice during the Clone Wars and I failed him: The Emperor, who is also a powerful Sith lord, turned him to the Dark Side. He betrayed and murdered the Jedi. But I think there’s still good in him, and it’s possible he could be turned back to the light side of the Force.

Hopefully it doesn’t require too much explanation to understand why this is not the most effective way of presenting this information.

Perhaps the most obvious one is that “Darth Vader is your dad” doesn’t really pack the same emotional punch when you just found out who Darth Vader was a couple sentences earlier.

But also note how this puts the procedural conclusion of the story — in which Luke tries to redeem his father — front and center from the first session of the campaign. This means that the campaign is either going to be very short, or it means that you’re going to keep hitting the “Luke confronts his father and tries to turn him to the light side” story beat over and over and over again until it becomes repetitive and boring.

THE THREE CLUE RULE

You may have noticed that I’ve been using the term “revelation” for each individual chunk of the exposition. This is a term I also use in the Three Clue Rule, and that’s not a coincidence. Each drip of exposition is a conclusion that you want the PCs to make and, therefore, the Three Clue Rule applies:

For any conclusion that you want the PCs to make, include at least three clues.

Take, for example, the conclusion that Darth Vader is a really bad guy. Lay out the clues:

• He attacks Leia’s ship at the beginning of the scenario and kidnaps her.
• Ben Kenobi tells Luke that Darth Vader murdered his father.
• Darth Vader kills Ben Kenobi.
• Darth Vader freezes Han Solo in carbonite.

(You might note that some of the other drip revelations mentioned above don’t have three separate clues appear in the films: That’s because they’re films, not an RPG scenario. Or, if we want to live inside our analogy, the films are a specific actual play and the PCs missed some of the clues. That’s why you follow the Three Clue Rule in the first place, right?)

Also keep in mind the corollaries of the Three Clue Rule, like using permissive clue-finding to opportunistically seize opportunities to establish revelations when PCs take unexpected actions.

DRIP DEVELOPMENT

As with mystery scenarios, one of the great things about implementing the Three Clue Rule is that it usually forces you to creatively engage with the material to a depth where the material sort of takes on a life of its own.

For example, as we begin designing this Darth Vader exposition drip we immediately run up against a conundrum: Why doesn’t Obi-Wan just tell Luke all of this stuff?

“He must be scared to tell him,” our hypothetical GM muses. “Obi-Wan lost Anakin. He was seduced by the dark side of the Force. Obi-Wan must worry that if Luke knew the truth about his father, then he would also be seduced. If Luke knew the truth, that strong emotional connection would become a conduit for his own corruption. So he lies. But he can’t conceal the truth entirely… he tells the truth from a certain point of view. The good man who was Anakin Skywalker was destroyed when he was consumed by the dark side; he was transformed into something else.”

In this process, our hypothetical GM has suddenly made the Force — which he’d previously just conceived of as kind of like good energy vs. demonic energy — into a more richly textured metaphysic. There’s now a whole concept of emotional relationships, and Darth Vader’s own struggle with identity has also emerged from this thought process.

You’ll find this sort of thing happening all the time when you design exposition drips: Either figuring out why the information is fractured in the first place will create interesting consequences; or the need to create a multitude of clues will create all kinds of texture you didn’t have previously; or in placing those clues you’ll find that you’ve created connections and reincorporated aspects of the game world in unanticipated ways.

INTERCONNECTED DRIPS

Wait… does Luke even know about the Force yet?

You’ll often find that certain revelations in an exposition drip also need a greater context to be meaningful. This can easily result in your needing to create another exposition drip, particularly if that greater context has significance elsewhere in the campaign. (As, for example, the Force does.)

We can also see that some of the revelations in the Force exposition drip need to be established before revelations on the Darth Vader drip, but not all of them. As a result, the two drips can be interwoven with each other throughout the campaign, with the PCs learning information about one drip and then the other. In fact, not only are these drips interwoven, but they frequently overlap — some revelations are shared between the exposition drips; or the same node of information can serve double duty by providing separate pieces of information to each drip.

The question of how many exposition drips you can have running simultaneously — how many balls you can keep in the air — is determined by the amount of complexity you and your players are capable of handling (and also, of course, what’s appropriate for the current scenario). But by having multiple exposition drips that connect to each other in different ways, you’ll usually end up with a totality that’s greater than the sum of its parts.

NON-LINEAR DRIPS

It’s easy to think about exposition drips in a linear fashion: The PCs learn A, then they learn B, then they learn C, and so forth. But as you look at each revelation in your exposition drip and consider the context necessary for that drip to exist, you’ll usually discover that this context does not, in fact, include many or any of the other revelations in the drip.

For example, you need to know that Anakin Skywalker was a hero during the Clone Wars before you discover that he was seduced by the dark side. (You need to know he was a hero before you can reveal that he’s a corrupted hero.)

But do you need to establish that Darth Vader is the Emperor’s right hand before establishing that Anakin Skywalker was a hero during the Clone Wars? Do you need to know that Anakin Skywalker is Darth Vader before discovering that there’s still good in Darth Vader?

Probably not.

So you’ll often discover that you DON’T need to worry about sequencing information: You can seed clues for a bunch of different revelations without worry about the specific order in which those clues are discovered. This is great news because (a) it makes it a lot easier to seed a node-based scenario with an exposition drip and (b) it really opens the door to discovering the unexpected during actual play.

(Take a moment to imagine what happens if Luke discovers that there’s still good in Darth Vader before he learns that Darth Vader is his father?)

NON-NARRATIVE DRIPS

As a final note here, the exposition drip technique isn’t necessarily limited to diegetic information in the game world. You can use a similar technique for mechanical concepts.

In the Eternal Lies campaign for Trail of Cthulhu, for example, the penultimate scenario features the PCs likely needing to lug explosives on a lengthy wilderness expedition. There are some mechanics involved with this which can lead to dramatic complications, but the first time I ran the campaign it was distracting trying to both come to grips with these mechanics and use them to good effect at a big, climactic moment.

So the second time I ran Eternal Lies, I found ways to incorporate these mechanics into the earlier scenarios in the campaign. As a result, by the time the PCs reached the final scenario, they were already familiar with thinking about explosives in terms of “charges” and figuring out how to deal with them on long expeditions.

This sort of thing is more esoteric than narrative exposition drips, but when you’re planning out a campaign take a look at the big, climactic moments and think about what novel mechanics are going to factor into those moments. Then track backwards and figure out how you can introduce those mechanics earlier in the campaign.

For example, if the big finale of this space opera campaign you’re planning has Luke and his father using the Dark/Light personality mechanics to try to sway each other’s metaphysical allegiance, try to figure out some moments earlier in the campaign where Luke can be tempted by the dark side so that both you and Luke’s player can become familiar with how those mechanics work. Maybe Yoda could send him to a dark side cyst on Dagobah? The Test of the Cave?

Yeah, that works.

Go to Part 1

No.

Bad GM.

No cookie.

Okay, we’ve been talking about things GM’s shouldn’t do for awhile now. So let’s talk about the elephant in the room: Fudging.

The most common form of fudging, and that from which the technique takes its name, is changing the outcome of a die roll: You fudge the result. If the die roll is done in secret, then you can just ignore it. If it’s done in the open, then you can invert the result by tweaking the modifiers involved. More advanced fudging methods can include stuff like adding extra hit points to a monster’s total in order to keep them alive.

But, regardless of the specifics, fudging is when a mechanical resolution tells you one thing and the GM chooses to ignore the rules and declare a different outcome.

JUSTIFICATIONS FOR FUDGING

Okay, let’s talk about the reasons GMs do this. All of these, of course, ultimately boil down to the GM not liking something that the resolution mechanics are telling them. The question is why the GM is unhappy with it.

#1 – Railroading. This one is pretty straightforward: Railroading happens when the GM negates a player’s choice in order to enforce a preconceived outcome. Enforcing failure (so that the PC can’t do what the player wants) is a really common way of railroading the game, and fudging is a really easy way to enforce failure.

See The Railroading Manifesto for a lengthy discussion of this topic and all the reasons why railroading is terrible and you should never do it.

#2 – To prevent a player character’s death. Or, in some cases, GMs will only fudge if it’s to prevent a total party kill — the death of ALL player characters. TPKs tend to kill campaigns (at least those not built around open tables), and lots of people would prefer to fudge the outcome of a fight (particularly if they feel that it’s just due to “bad luck” or whatever).

See The TPK Gamble for a specific discussion of this.

#3 – To make the story “better.” The most infamous version of this is, “But they can’t kill the Big Bad Guy now! He’s supposed to survive to Act III!”

I say infamous for good reason here: Players hate this shit with the fiery passion of a thousand burning suns. And you basically can’t throw a stone in RPG circles without hitting someone who has a story about the time their GM pissed them off by doing it. Check out The Principles of RPG Villainy for a better alternative.

#4 – To correct a mistake you’ve made. Maybe you’ve been screwing up a mechanic for the whole fight and it’s made things much harder for the PCs than it should have been. Or you accidentally doubled the number of guards when the fight started. Or, going even further back, maybe you just screwed up the encounter design and something that should have been easy for the PCs is actually incredibly difficult. So you fudge something to bring it back in line with what it was supposed to be or should have been.

This is actually pretty understandable, and I discuss the difference between openly retconning a mistake and silently retconning a mistake in Whoops, Forgot the Wolf. But you can easily find yourself slipping from “fixing a screw-up” to “enforcing a preconceived outcome” here and end up back in railroading. So use caution.

DON’T FUDGE

In the end, all fudging is the GM overriding a mechanical outcome and creating a different outcome which they believe to be preferable (for whatever reason).

Over the thirty years I’ve been doing this, however, I’ve learned that many of the most memorable experiences at the table are the result of the dice taking you places that you never could have anticipated going. Fudging kills those experiences.

But what if the mechanical outcome really is terrible and would make both you and your players miserable?

If you and/or your players truly can’t live with the outcome of a dice roll, then you made a mistake by rolling the dice in the first place. You need to focus on fixing that problem.

This applies beyond individual dice rolls, too. If you don’t want the PCs to die, for example, why are you framing scenes in which death is what’s at stake? (This is a rhetorical question: GMs do this because D&D teaches them to (a) frame lots of combat scenes and (b) make the default stakes of any combat scene death.)

The Art of Pacing talks about the scene’s agenda being the question which the scene is designed to answer. (For example, “Can Donna convince Danny to go into rehab?”) If the question is, “Will the PCs die?” and the answer is always, “Absolutely not.” then the scene is drained of meaning and becomes a boring exercise.

This is why, when the players figure out that the GM is fudging (and they will), it deflates tension and robs them of a legitimate sense of accomplishment. What was once meaningful is suddenly revealed to be meaningless. And this is the biggest problem with fudging: It may fix an immediate problem, but it will inflict permanent damage on everything.

In a very real sense, fudging is a betrayal of trust. And once you, as the GM, lose the players’ trust, it becomes virtually impossible to regain it. Fudging ends up tainting everything you do: It removes the real magic of an RPG campaign and turns it into a cheap magic trick. Once the players spot the trick (and, again, they will), the magic vanishes entirely and you’re left with a hollow experience.

Regaining their trust and making them believe in the magic again is really difficult.

TRIAGE AT THE TABLE

Here’s my controversial rule of thumb:

The more you fudge, the shittier you are as a GM – either because you are fudging or because you need to.

If you’re not just fudging to be an asshole and screw over your players, then you’re ultimately fudging in order to fix something that has gone wrong:

• You adjudicated the resolution poorly.
• You designed the scenario badly.
• You screwed something up and need to correct it.
• You’re using a set of rules which creates results you and/or your players aren’t happy with.

And so forth.

This is not to say that you should never fudge. Mistakes happen and we don’t need to live with those mistakes in the pursuit of some unrealistic ideal. But every time you do fudge, you should view that as a failure and try to figure out how you can fix the underlying problem instead of just continuing to suck in perpetuity:

• Don’t roll the dice if you can’t live with the outcome. (And, ideally, learn how to still create meaningful stakes instead of just skipping the resolution entirely.)
• Figure out how to design robust scenarios that don’t break while you’re running them.
• Create house rules to permanently fix mechanics that are creating undesired results. Or, if the system is completely out of line with what you and your players want, swap to a different system.

And so forth.

Next time you find yourself in a position during the game where you feel it’s necessary to fudge, I want you to do a couple of things.

First, ask yourself: Is it truly necessary to fudge in this moment? Is it necessary to reject the improvisation prompt of the mechanical resolution’s outcome, or can you find a way to work with that outcome to create something interesting and enjoyable? At the stage in the resolution process where you’re narrating outcome, you usually still have a lot of power as the GM. An easy example of this is failing forward: Instead of the PC failing in what they wanted to do, they succeed with a negative twist or consequence.

But also, to a certain extent, just take a moment to second guess yourself: The outcome which you initially think cannot possibly happen, often can happen. It’s just not what you expected or would have done of your own volition. Try to push back that initial moment of rejection and really, truly think about what the outcome would be and whether there’s interesting and cool stuff that lies beyond that outcome.

Second, ask yourself: Can I just be open and honest with my players in this moment? Instead of secretly fudging the outcome, could you just explain to the players that, for example, you screwed up the encounter and things need to be retconned a bit?

And maybe you can’t! There are circumstances where you’re better off plastering over the cracks of your mistake with a cheap magic trick instead of damaging the players’ immediate immersion and engagement with the game world. It’s not ideal, but sometimes that’s the best you can do for right now. You’ll just have to learn from your mistakes and do better next time.

CODA

If you’re still a proponent of fudging, let me ask you a final question: Would you be okay with your players fudging their die rolls and stats and hit point totals?

If not, why not?

If you truly believe that fudging is necessary in order for you to preserve the enjoyment of the entire table, why do you feel you know better than the other people at the table what they would enjoy?

Think about it.

The Fudging Corollary: Not All Dice Rolls Are Mechanics

Go to Part 10: Idea Rolls

SESSION 20F: THE GHOST APPEARS

April 27th, 2008
The 9th Day of Kadal in the 790th Year of the Seyrunian Dynasty

Heading back out into the hallway they went to the last door on the second level. This was another iron door and it led into one of the small towers that flanked the front gate. A ladder bolted to the wall led up to the next level of the tower.

Tee and Tor climbed up the ladder. Tor headed through another iron door, this one leading to the gatehouse immediately above the entrance to the house: They could see where a large stone block had been levered out of the floor and pushed to one side, revealing the murder hole the ratlings had attacked them through.  A narrow wooden table off to one side held the decrepit remains of four crossbows and three quivers of rotten quarrels, all covered with cobwebs and dust. An iron pot filled to the brim with rusty caltrops was shoved into a far corner. There was a matching door directly opposite.

Tor proceeded cautiously into the gatehouse. He hadn’t gone more than a few steps, however, before the door suddenly slammed shut behind him. Tee jumped for it and easily got it open again. She turned and called over her shoulder, “Get up here! Something’s happening!”

The trapdoor slammed shut.

“Tee?” Elestra called. “What’s happening?”

Tee whirled back towards Tor… just in time to see the ghost materialize between them.

The spirit wore the robes of an Imperial priest, but its face was contorted with fury. “Leave this place! The curse will claim your souls!”

Tee hesitated for a moment and then leapt for the trap door, yanking it open. “Agnarr! The ghost is right here!”

Tor, meanwhile, had drawn his sword and – with a single quick swing – sliced it through the ghost’s ethereal form. Although the blade crackled and its electrical arcs flashed as it passed through the ghost, the apparition appeared unphased.

Agnarr began clambering up the tower ladder. Dominic, thinking quickly, ran back around the hall to a window looking out over the courtyard. Through this he was able to look up through one of the inner arrow slits of the gatehouse and see the ghost moving menacingly towards Tor.

Dominic raised his holy symbol and called out a prayer to Athor. But whether it was the distance, the thick stone walls, or the sheer tenacity of the spirit the prayer had no effect. Frowning, Dominic ran back around towards the ladder.

Tor swung his sword again… again to little effect. But at the blow the ghost’s face was transformed into a black maw of rage “YOU HAVE BEEN WARNED!”

Every object in the gatehouse began to shake violently, and then handfuls of the sharp, rusty caltrops came flying out of their cauldron – pelting Tor viciously.

Agnarr leapt out of the trapdoor and drew his sword, bounding towards the door leading to the gatehouse. “FOR THE GLORY!”

The spirit whirled: “LEAVE THIS PLACE.”

Agnarr grunted and swung his flaming sword. It ripped through the ghost, and Agnarr could feel it catching and tearing.

The ghost moaned in pain and rushed away from Agnarr… passing straight into Tor’s body.

Tor jerked spasmodically, and then a clearly alien intellect took possession of his limbs and spoke through his lips: “Leave this place or your friend will die.”

Agnarr paused. “I’ll only give you once chance: Get out of his body.”

“LEAVE THIS PLACE!”

Agnarr attacked. The spirit clumsily raised Tor’s sword and parried the attack. Agnarr moved to attack again, but the ruined crossbows were swept off their table and hurled at Agnarr by invisible hands.

Agnarr stumbled under the assault, and barely got his sword back into a defensive position as “Tor” attacked him. Agnarr parried several more attacks, trying to figure out some way of getting rid of the ghost without harming Tor. But there didn’t seem to be any way around it.

“I’m sorry, Tor! Dominic will heal you later!” Agnarr got ready to swing away with all his strength, which would surely sweep aside the ghost’s clumsy defense—

When Dominic, having ascended the ladder behind him, raised his holy symbol and with a shouted prayer focused his faith upon Tor’s body. The ghost was blasted back, forcibly ripped from Tor’s soul, and then faded into wispy nothingness…

“Is it gone?” Tee asked.

Dominic gasped. “I think so.” (more…)