4
\$\begingroup\$

A Shield spell is cast as a reaction to an attack that hits. It applies its AC bonus even against the attack to which it is a reaction, meaning that it can make that attack that hit, retroactively miss and thereby not do damage.

Shield 'interrupts'/potentially cancels its trigger, as stated in the DMG in the 'Adjudicating Reaction Timing' section.

Since the damage has retroactively not been done, what happens to effects that were triggered by the damage in the first place? Are they undone as well, or do they persist?

This question has two answers of nearly equal popularity, with one arguing that other effects triggered by the hit still persist, and the other arguing that they do not. On the side of the argument that they do not occur is a (now-unofficial) tweet from Jeremy Crawford stating

If the attack has a special effect that relies on it hitting, that effect doesn't occur if the attack is turned into a miss.

If we accept that Shield means the effects of a hit retroactively do not happen, what occurs when those effects include ones that allowed the Shield spell to be cast in the first place?

I am currently running Curse of Strahd and in today's session the party was fighting flameskulls, which are undead capable of casting the Shield spell.

  1. The party cleric had turned multiple flameskulls.
    "Channel Divinity: Turn Undead":

A turned creature must spend its turns trying to move as far away from you as it can, and it can't willingly move to a space within 30 feet of you. It also can't take Reactions.

At the start of the paladin's turn, the flameskull could not cast Shield, since it was Turned, and could thus not use reactions.

  1. The party paladin then attacked a flameskull that had been turned, and hit, causing damage.
  2. This damage removed the turning effect on the flameskull, allowing it to cast spells.
    "Channel Divinity: Turn Undead"

it is turned for 1 minute or until it takes any damage.

Once the flameskull had taken damage, it was no longer Turned. Since it was no longer Turned, it was able to use its reaction, and thus able to cast Shield.

  1. As a reaction, the flameskull cast Shield, which resulted in the effect that it was retroactively not hit.

  2. Since it had not been hit, it had not taken damage. What happens next?

Option A: Although the damage as a consequence of the hit is removed, the fact that the damage was done at one point in time was enough. The spell slot for Shield is removed, the flameskull is undamaged but no longer turned, and it finishes its turn.

Option B: Since the damage as a consequence of the hit is removed, the flameskull is retroactively still turned. The spell slot for Shield is removed, but the flameskull is still Turned, and finishes its turn.

Option C: Since the damage as a consequence of the hit is removed, the flameskull is retroactively still turned. The spell slot for Shield is removed. Since the flameskull is retroactively still Turned, it could not have taken a reaction and thus did not cast the Shield spell. The spell slot for the Shield spell is restored. Since it did not cast Shield, the flameskull was actually hit and took damage. Since it was damaged, it is no longer turned and can now cast Shield. The flameskull never finishes its turn because it is caught in an infinite recursion loop.

\$\endgroup\$
20
\$\begingroup\$

Shield can only be cast during the resolution of an attack, not after

The resolution of an attack is described thusly:

3. Resolve the attack. You make the attack roll. On a hit, you roll damage, unless the particular attack has rules that specify otherwise. Some attacks cause special effects in addition to or instead of damage.

The resolution step is effectively 2 steps: first determine if the attack hit, then resolve the effects of a hit or miss. Importantly, the proper timing for casting Shield is before the resolution of the attack's effects, while you are still determining whether the attack hits. (Obviously, in practice it is common for to declare the hit and damage at the same time for the sake of expediency and still allow casting Shield afterward, but logically speaking, they are separate steps.)

At the time the DM declares that the attack hits the flameskull, its reaction is not available, because it is still turned until the attack's effects are resolved. This would be the window of opportunity to cast Shield if it could. But since it cannot do so, the attack resolves normally, and the flameskull suffers all the effects of a hit, including ending the effect of Turn Undead. At this point, the attack has already been resolved, and casting Shield is no longer an option. In summary: the flameskull is hit by the attack, and no spell slot is expended because no shield spell is cast.

Narrative justification

You might not like the above logic. For example, the accepted answer to this question, which seems pretty reasonable, would imply a different conclusion to the situation proposed in the current question. However, in the case of Shield, I think we can also give a narrative justificaion for why the flameskull can't cast shield against the attack that ends the effect of Turn Undead. In terms of the game mechanics, the shield spell works kind of like an "undo" button, forcing you to go back and re-resolve a previously resolved result (the hit). However, unless you and your DM have agreed that your shield spell literally involves time travel (e.g. a Dunamancy-based variant of the spell), then the mechanics don't match how the spell is typically described in narrative terms. Most commonly, shield is described as being cast "just in time" before the attack to prevent it from hitting you.

If this is how you describe the spell, then logically the flameskull cannot cast shield while turned. On the other hand, if in your game shield literally involves time travel, then it might make perfect sense to allow it to work in this case. The apparent inconsistency is resolved by the time reversal that happens when the spell is cast. This is an instance of the DM filling in the gaps in the rules, as described in the DMG's introduction:

The rules don’t account for every possible situation that might arise during a typical D&D session. For example, a player might want his or her character to hurl a brazier full of hot coals into a monster’s face. How you determine the outcome of this action is up to you.

In this case, I'm suggesting that you use the in-universe description of how the spell works to decide whether and how it works in a specific situation that is not (adequately) covered by the rules as written.

\$\endgroup\$
5
\$\begingroup\$

In order for the flameskull to be able to take a reaction, it has to have received damage, so there is no reaction to take

Resolving an attack has three steps:

  1. Choose a target. [...]
  2. Determine modifiers. [...]
  3. Resolve the attack. You make the attack roll. On a hit, you roll damage, unless the particular attack has rules that specify otherwise. Some attacks cause special effects in addition to or instead of damage.

So the order of operations is:

  1. Make an attack on the target with the appropriate modifiers
  2. Determine whether or not the attack hits.
  3. If there is a hit, deal damage.

The PHB, describes reactions as (emphasis mine):

A reaction is an instant response to a trigger of some kind, which can occur on your turn or someone else's.

Tasha's Cauldron of Everything nicely restates the rules on reaction timings (available in the DMG) in a succinct way:

If you're unsure when a reaction occurs in relation to its trigger, here's the rule: the reaction happens after its trigger, unless the description of the reaction explicitly says otherwise [...]

There is one element from the more verbose DMG rules that are relevant here:

[...] For example, the opportunity attack and the shield spell are clear about the fact that they can interrupt their triggers. [...]

So, shield is called out as a specific exception to the normal rules, in that it can interrupt its trigger.

The trigger for the Shield spell is:

when you are hit by an attack or targeted by the magic missile spell.

Using these rules, and applying it to this order of operations, the reaction for shield happens in between operations 2 & 3 since shield is an instant response to the hit (ie it happens before we determine if damage is applied). In particular, the result of the shield reaction is that it can modify the outcome of operation 2 in our order of operations. Damage can only occur if the result, after operation 2 has completed, including the shield modifications, is a hit.

While a creature is under the effect of turn undead, it cannot take reactions:

[...] It also can't take Reactions.
Turn undead, as you noted, only ends on the target if:

[...] it is turned for 1 minute or until it takes any damage.

Thus, it cannot respond to the trigger (being hit) with the shield spell, as it does not have reactions at that point. Furthermore, it cannot use its reaction after the damage has resolved, as other events have resolved (ie damage) after the trigger.

If the reaction could be taken after the damage occurs, then it is no longer an instant response to the trigger, instead it becomes a delayed response (as some events have passed between it an the trigger). This is clearly a non-sensical route to take, because if we allow reactions to be delayed responses, then multiple events can happen between the trigger and its resolution, allowing players and NPCs to "store up" reaction triggers.

So what happens with the flameskull?

The paladin hits the flameskull, deals damage to it, and the flameskull is released from Turn Undead. If any other character (or indeed the Paladin) attacks the Flameskull again before its next turn, then the flameskull now has it's reaction available to cast shield, if necessary, against those subsequent attacks.

\$\endgroup\$
  • \$\begingroup\$ "Furthermore, it cannot use its reaction after the damage has resolved, as other events have resolved (ie damage) after the trigger. If the reaction could be taken after the damage occurs, then it is no longer an instant response to the trigger, instead it becomes a delayed response (as some events have passed between it an the trigger)." This makes sense to me, and is keeping with the WoTC definition of "instant" as established in MTG. Is this logical principle anywhere explicitly stated in rules? It would be useful in dealing with other questions I have seen. \$\endgroup\$ – Kirt Nov 18 '20 at 16:03
  • 1
    \$\begingroup\$ @Kirt The definitions from MtG do not apply here at all. \$\endgroup\$ – Mark Wells Nov 18 '20 at 17:43
  • \$\begingroup\$ @MarkWells Which is why I think this answer would be stronger with a rules citation rather than just a logical argument. \$\endgroup\$ – Kirt Nov 18 '20 at 18:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.