The potion intuition everyone carries — "rares are rarer" — is true on exactly one drop path and false everywhere else.
The roll everyone misreads
StS1's potion rarity roll is 65/25/10 (under 65 COMMON, under 90 UNCOMMON, else RARE) — but that roll only governs reward-screen potions, where the game rerolls until the rolled rarity matches. The actual combat-end or event drop is drawn uniformly from the run's potion pool: 30 shared + 3 class-exclusive keys (33 obtainable) picked with one even roll. Rarity labels don't weight that draw at all: every one of the 33 potions is equally likely per pick. Rare potions really are rarer from rewards, and nowhere else.
Slots and multipliers
Base 3 slots; A11 takes one away (the ladder beat mirrored by StS2's A4 belt modifier); Potion Belt adds two. Sacred Bark doubles every potion effect via a relic hook. Shop prices stay 50/75/100 by rarity in both games.
The placeholder that explained row 43
StS1 harvests 43 potion classes, but only 42 are potions: the Potion Slot
key is the belt's own filler UI class, counted so the belt keeps its name.
Recording it as an explicit placeholder (not dropping it) is why the canonical
count stays clean.
StS2's flat design
StS2 holds 64 potions in a perfect 20/20/20 Common/Uncommon/Rare split, plus 2 Event, 1 Token, 1 unclassified internal row. Rows carry typed usage and target fields instead of StS1's thrown/consumed flags, and shop pricing keeps the same 50/75/100 table as StS1 — one of the few constants the sequel didn't touch.