Spell Creation

While the AD&D Dungeon Master's Guide gave guidelines for allowing a character to create new spells, the current edition says little about the activity. These are my recommendations on the subject. An entire book could be written on this subject, and it would still have to be somewhat incomplete, because every realm is different, and every campaign is different.

A world with several independant colleges devoted to magic, supported by sprawling libraries, would easily have hundreds, if not thousands, of unique spells. "Discovery" of a new spell would likely consist of pouring over dozens of spell indexes in an attempt to find a spell that was once created, but was somehow forgotten. Or perhaps it would entail finding the research notes of several mages who attempted to create such a spell, combining and reworking old theroms, and then finally succeeding (or leaving your notes and studies for the next hapless asperant). Coming up with a truely novel and unique spell idea in such a realm would be exceptional indeed.

In a world where magic was indeed rare, the formation of a unique idea for a spell would be relativly easy, but finding the failed attempts of others would be much harder, if such research even exists. Creation of a new spell would be wholely independant work, based on extending the base understandings of magic beyond where it once existed.

I recommend any DM who considers allowing a character to create a spell, they should first consider their campaign, and determine where a proposed spell would exist; would such a spell likely exist somewhere in your world, or is the spell of such a nature that no wizard in the entire history of your world would have ever thought to create it?

Stages of a Spell's Life

It is probably easiest to think of a spell (and the related spell formulae) as an IDEA. Ideas form, come into existance, and persist until a new, better idea replaces it (a meme). No one researches firecube, because fireball exists, and it works just fine.

1. Entirely unknown, never even considered
2. Theoritical spell
3. Research
4. Discovery!
5. Added to a spellbook somewhere
6. Shared with apprentices, fellow casters, etc.
7. Makes it way into a library
8. Taught as a special "named" spell
9. Becomes widely known
10. Becomes universally known

Up until the fourth step, the spell doesn't exist as a castable spell. But once it has been discovered/created, the formulae that describes that spell and it's effects is known to it's creator, so that creator can cast the spell. The spell may not be shared, but it is likely that at least some research notes were created, as well as at least one copy in the wizard's personal spellbook. Unless great care was taken by a creator in the past, the research notes will persist somewhere, and even if the spell was never shared, rumors from people seeing the spell cast will live on in legends, old stories, or entries in a journal or diary.

So assuming the knowledge of a spell isn't lost with the death of it's creator, the spell will eventually become known among those who make a special study of magic (such as wizards and sages). These are the spells of Leomond and Tenser, the spells of Tasha, Drawmij, and Mordenkainen. Some of these spells become as well-known and widely used as Identify, Shield, and Fireball.

While powerful spells are well-known, what wizard remembers the flame-based spell that Fireball replaced? Or did such a spell ever exist? Allowing characters to discover ancient spells that were replaced by better ones is one thing you might add for the researcher (Perhaps flameball; a 3rd level evocation spell that creates a 15-foot-radius, 6d6 fire damage ball, with a 180-foot range, that cannot be upcast. It has a slighly better range, but isn't as big or powerful. It could also require a glowing coal as a material component (something not as easily carried around)).

The Numbers

Assuming the player has a spell description that has been approved by the DM, the character may begin work. Assuming the character has access to an adequate library, shrine, or temple as appropriate; they must spend:
200 gp / spell level / week of work
for a number of weeks equal to the spell level. These are the initial costs; gathering a varitity of spell components, purchasing tools and materials necessary, hiring the services of sages and researchers, etc. After the initial time and gold expendatures, the player may spend a week of work, and then make a successful Arcana check:
DC = 10 + 2* spell level
The DC decreases by one for each week of work, to a minimum of (10 + spell level).

If the character does not have access to an appropriate library, they must spend up to ten times (2,000 gp / spell level / week of work) the amount listed. This gold is spent on aquiring a library, constructing/improving their shrine, etc. A character may spend less (as much as half) by curating a special library, building a specially dedicated shrine, etc., although doing so will only benefit the spell currently being researched, and may be of little to no benefit in researching a different spell.