A few years back these posts over on Ode to Black Dougal pointed out that B/X spell books work differently than spell books in other D&D variants. In brief, B/X magic-users cannot add spells to their spell book from scrolls and the capacity of their spell books is limited to the number of spells they can actually cast. For example, a level 5 magic-user could have at most 2/2/1 spells in their spell book, that's it. Said magic-user could still memorize a spell multiple times of course, so if their book contained Read Magic and Sleep they could memorize either Read Magic/Read Magic or Read Magic/Sleep or Sleep/Sleep. What they cannot do is add another first-level spell to their spell book, that's that "sweet treat" reserved for magic-users of level 7 or higher.
Apparently I am not the only one intruiged by this so-called "strict reading" of the magic rules in B/X: At least Alex Schroeder seems to agree, but undoubtedly there must be a few more of us? (Feel free to give a shout-out in the comments!) I am actually a recent convert to this approach, for the longest time I thought that BECMI and AD&D had it right instead. Truth be told, I didn't even realize that B/X was different until I saw those posts: The BECMI approach was so deeply ingrained that I couldn't read the B/X rules correctly!
So why do I like this "strict reading" of the magic rules? After all, it seems to significantly "nerf" the poor magic-user whereas more recent editions of D&D have done almost the exact opposite. Let me take you through it...
First and foremost I like that it makes learning a new spell feel more like a significant life choice and less like just adding another stamp to Ye Olde Collection. Under BECMI rules a character who finds two scrolls does not have to make a choice at all: They can simply add both spells to their spell book, case closed. Under AD&D rules there might be a choice, for magic-users with a low intelligence score anyway. On page 10 of the Player's Handbook we learn that magic-users with an intelligence of 9 have at most 6 spells per level in their spell books. However, those with an intelligence of 18 are allowed 18 spells per level and those with an intelligence of 19 have no limit whatsoever. Quickly, when was the last time you saw an AD&D magic-user with a low intelligence score? (Granted, Unearthed Arcana piles some rules about the capacity of spell books on top of that, but those really just turn high-level magic-users into caravans of wagons and guards, traveling libraries if you will. Or maybe into "portable hole aficionados" for that matter.) Ignoring for the moment that spells can't be copied from scrolls, the limits in B/X are much lower than the limits in AD&D: A level 14 magic-user only has "room" for 4/4/4/4/3/3 spells! So suddenly it's not clear anymore that they should really learn Fireball and Lightning Bolt. Or look at it the other way around: A magic-user who does learn both of these spells has obviously chosen the path of a "war wizard" of sorts.
A second reason for preferring the "strict reading" in B/X is that it encourages what I would consider "wizardly" behavior. You cannot just "hunt for scrolls" in monster lairs! Instead you have to either develop "good professional relationships" to potential teachers or you have to lock yourself away for weeks at a time doing "spell research" on your own. Of course there are also "darker" options like blackmailing a potential teacher or even a "demonic pact" of some sort. Whatever it is, there are more interesting choices to be made and more depth is added to the campaign: Cabals of wizards make more sense, a magic-user's reputation suddenly matters, potential teachers may have quests that need completing, etc.
The nature of spell scrolls provides a third and (for now) final reason. In all D&D variants, magic-users who have a scroll and have read it are able to cast the spell from the scroll, thereby erasing it. In AD&D we're told explicitly (see Player's Handbook, page 100) that casting from a scroll only requires speaking the magical incantation, it does not require somatic or material components. It stands to reason that the magic-user who scribed the scroll had to somehow "encode" those things into the scroll thus making them unnecessary for casting from the scroll. This is further coroborated (see Dungeon Master's Guide, page 13) when we're told that spells that normally age the caster actually age the magic-user scribing the scroll, not the magic-user casting from the scroll. In other words, scrolls somehow contain the magical energies of a partially cast spell, lacking only the verbal component (which functions as some kind of "trigger" for releasing those energies). This understanding of scrolls can be used to explain why scrolls cannot be scribed into spell books: The "actual" description of the spell must contain details about somatic and material components and since those are missing from the scroll the information required to cast the spell "from scratch" is incomplete. (Note that of course AD&D goes on to say that you can copy from a scroll into a spell book anyway, but luckily we're not trying to understand AD&D, we're just trying to understand B/X. AD&D actually makes things even worse by allowing magic-users to cast directly from their spell books, see Unearthed Arcana, page 80...) Needless to say I find this distinction between "partially cast spells" on scrolls and "complete spells" in spell books quite satisfying. Gary would probably scold me for trying to use some kind of "consistent reasoning" in a game that features dragons and spells in the first place, but I still prefer for things to "make sense" when they can.
So there you have it, my interpretation of and reasoning about the "strict reading" of the B/X magic system. But I am not quite done. I can never leave "good enough" alone. The strange compulsion to "house rule" drives me further.
Quick, in B/X D&D, what's the difference between a magic-user with intelligence 9 and a magic-user with intelligence 18? That's right, the difference is three more languages and an XP bonus of +10% (see Basic Rules, page B7). Being familiar with AD&D this has always bothered me about B/X, I felt like there should be some other mechanical effect to distinguish those two magic-users. However, things like "bonus spells" of the sort that AD&D gives to clerics for high wisdom scores never sat right with me. (For example it's strange that a high wisdom score only gives bonus spells of the first four levels (see Player's Handbook, page 11), but giving a certain number of bonus spells on each level seems too overpowered.)
Luckily the "strict reading" provides a new option: Have intelligence affect the number of spells/level a magic-user can have in their spell book! Since B/X (strictly speaking) doesn't have standardized modifiers (they are a BECMI invention) we use the number of additional languages and reinterpret it as "number of additional spells per level" that can be written into the spell book. So our level 5 magic-user from above can still cast at most 2/2/1 spells per day, but if he or she has an intelligence of 16 they could have 4/4/3 spells in their spell book (instead of just 2/2/1). The effect of intelligence on "Maximum Number of Spells/Level" from AD&D actually provides a precendent for this and the rule is simpler than what ACKS does with "repertoires" and the intelligence bonus for those (see Adventurer Conqueror King, page 66+). I don't think it's overpowered either, after all a level 14 magic-user with INT 18 could have only 7/7/7/7/6/6 in their spell book, a far cry from the 12/12/12/12/12/12 actually available in B/X.
Total win in my book.
Addendum: While writing all of this up I noticed that B/X interprets low intelligence scores as a limited ability to speak/read/write. However, magic-users are not required to have a minimum intelligence score. This either implies magic-users who can be almost braindead yet cast Fireball or a missing minimum requirement. I am about to house-rule that magic-users need a minimum intelligence of 9 just like elves. Whether I will allow elves to have additional spells in their spell books for high intelligence also remains to be seen.
Update 2014/12/11: I just noticed that my friend Delta had already been over the "Where are the somatic components?" territory a few years ago, albeit in a slightly different context. Then again, Dan has written about everything D&D already, so maybe I don't have to feel too horribly bad...