Two percentages share this lesson, and they answer completely different questions. One grades a process. The other describes a molecule and is the same in every batch ever made.

Percent yield: $\%\,\text{yield} = \dfrac{m_{\text{actual}}}{m_{\text{theoretical}}} \times 100\%$, read aloud percent yield equals m-actual over m-theoretical, times a hundred. $m_{\text{actual}}$ is the mass you really isolated — what the bagging scale read. $m_{\text{theoretical}}$ is the mass stoichiometry allowed from the charge, computed on the sheet before the reactor was ever filled. It is a ceiling, not a target: no real batch reaches it, because filter cake holds product, mother liquor carries some away, and the last rinse takes a little more. A yield above 100% is not a triumph, it is a diagnosis — most often product that has not finished drying.

Two mistakes are worth naming before you meet them. Inverting it puts every batch above 100%. And computing $\dfrac{m_{\text{theoretical}} - m_{\text{actual}}}{m_{\text{theoretical}}}$ gives the percent lost — a genuinely useful number, and the exact complement of the one you were asked for. The two add to 100, so reporting one as the other turns a good batch into a terrible one on paper.

Percent composition: $\%X = \dfrac{a\,M_X}{M_{\text{compound}}} \times 100\%$. Here $X$ is the element you are asking about, $a$ is the number of its atoms in one formula unit — the subscript, a bare count: $M_X$ is that element's molar mass and $M_{\text{compound}}$ the molar mass of the whole compound. Note what it does not need: a sample. This is answered from the formula alone, which is exactly why a specification can be written in it.

The dropped subscript is the mistake this relation is famous for. Calcium carbonate is 40% calcium and 48% oxygen — and the oxygen figure only comes out right if you remember there are three of them. Count one and you under-report by two thirds.