Dimensional analysis is the habit of treating a unit as part of the arithmetic instead of as a label added at the end. Multiply by ratios that equal one, arrange them so the units you do not want cancel, and the units left standing tell you whether the chain was built correctly before you check the number.
Said without symbols: a unit is an instruction about what may be multiplied by what. OpenStax states the same principle in one sentence, that the units of quantities must be subjected to the same mathematical operations as their associated numbers, and everything below follows from taking that literally.
What is a conversion factor really?
A ratio that equals one. OpenStax defines it as a ratio of two equivalent quantities expressed with different measurement units, and the word equivalent is doing all the work. Because 1 L and 1000 mL are the same amount of stuff, the ratio 1000 mL / 1 L has the value one, and multiplying by one changes the units of a quantity without changing the quantity.
That has an immediate consequence worth saying out loud. Every equality gives you two conversion factors, and they are reciprocals. From 1 mol H2O = 18.015 g H2O you get both
18.015 g / 1 mol and 1 mol / 18.015 g
Neither is the right one in general. The right one is whichever puts the unit you are trying to get rid of on the bottom, so it cancels against the unit you already have. That is the entire decision, made fresh at every step, and it is why this method does not require remembering whether to multiply or divide.
Why does unit cancellation work as an error check?
Because a unit that fails to cancel is evidence that a factor went in upside down, and it is evidence you get before the answer, not after.
Suppose you want moles of water from 12.0 g of water and you reach for the wrong factor:
12.0 g x (18.015 g / 1 mol) = 216.18 g^2/mol
Nothing in that result is a quantity chemistry has a name for. Grams squared per mole is not a small error, it is a sign that the chain is backward. Flip the factor:
12.0 g x (1 mol / 18.015 g) = 0.666112 mol
Grams cancel, moles remain, and the number is now worth reading. Writing units on every line is what makes this check available, which is why it is worth the extra pen strokes even when the conversion feels obvious.
Which conversion factors are specific to chemistry?
This is where most explanations stop and most students stall. Metric prefixes are the easy half. The factors that actually carry a chemistry problem come from chemical facts rather than from definitions of units.
Molar mass, in grams per mole. The bridge between what a balance reads and what an equation counts. Its value comes from the IUPAC standard atomic weights, so it is looked up rather than recalled. For water, 2(1.0080) + 15.999 = 18.015 g/mol.
The Avogadro constant, in particles per mole. NIST gives it as 6.02214076 x 10^23 per mole, exact by the SI definition of the mole. This is the factor that turns an amount into a count of molecules.
Molarity, in moles per liter. A concentration is a conversion factor between a volume of solution and an amount of solute, which is how OpenStax uses it. It is moles per liter of solution, not of solvent.
Density, in grams per milliliter. Turns a measured volume of a liquid into a mass, which is the usual first step when a problem hands you milliliters of something that is not a solution.
The mole ratio, in moles per mole. This one is different in kind, because it comes from a balanced chemical equation rather than from a table. OpenStax describes the coefficients as deriving the stoichiometric factor that relates the amounts of two substances, and it is applied the same way any other factor is. Together with the Avogadro constant it is one of only two exact factors in this list, and the only one that is exact because it counts things rather than because it was defined.
Molar volume, in liters per mole. Available only for a gas, only under stated conditions, and better derived from the ideal gas law than quoted, because a quoted molar volume silently carries a temperature and a pressure with it.
What does a chain look like when it is written properly?
One line, factors in sequence, each one chosen so the previous unit cancels.
How many molecules of water are in 12.0 g of water?
N = 12.0 g x (1 mol / 18.015 g) x (6.02214076 x 10^23 molecules / 1 mol)
Grams cancel against the first factor, moles cancel against the second, molecules remain. The arithmetic gives 4.0114 x 10^23, reported as 4.01 x 10^23 molecules to three significant figures, since 12.0 g carries three and both factors are more precise than that.
What mass of sodium hydroxide is in 250.0 mL of 0.125 M solution?
m = 0.2500 L x (0.125 mol / 1 L) x (39.997 g / 1 mol)
Liters cancel, moles cancel, grams remain. That is 0.031250 mol and then 1.2499 g, reported as 1.25 g.
Notice that neither line required deciding to multiply or divide. Both required deciding which way up to write a factor, which is a smaller and much more mechanical question.
Where does a chain get long, and how do you keep it straight?
In a reaction problem, where the middle of the chain crosses the balanced equation. The pattern is always the same three moves: get into moles, cross using the mole ratio, get out of moles into whatever unit the question asked for.
25.0 mL of ethanol, whose density is 0.789 g/mL, is how many moles?
n = 25.0 mL x (0.789 g / 1 mL) x (1 mol / 46.069 g) = 0.428162 mol, reported as 0.428 mol
The molar mass of ethanol, C2H5OH, is 2(12.011) + 6(1.0080) + 15.999 = 46.069 g/mol. Milliliters cancel, grams cancel, moles remain. Adding a mole ratio and a second molar mass on the end of that line would answer a full stoichiometry question, which is exactly what the worked practice set does four times over, and what the five-step method formalizes.
Two habits keep a long chain honest. Write it as one continuous line rather than as a stack of separate small answers, so nothing gets rounded in the middle. And read the units left to right before evaluating anything, confirming that every unit except the final one appears once on top and once on the bottom.
Where does dimensional analysis stop working?
At any relationship that is not a simple proportion, and temperature is the case everyone meets first.
OpenStax is explicit about why. Most units are directly proportional, of the form y = mx, so a single ratio converts between them. The Celsius and Fahrenheit scales do not share a common zero point, so converting between them needs the form y = mx + b, with an offset. A ratio of 9 degrees Fahrenheit per 5 degrees Celsius does exist, and OpenStax uses it as the slope, but it converts a temperature difference rather than a temperature reading. A reading also needs the offset, so no single factor can be canceled against it.
Kelvin and Celsius are the same size of degree but also do not share a zero, which is why T(K) = T(degrees C) + 273.15 is an addition and never a multiplication. This matters constantly in gas problems, where a temperature has to be in kelvin before PV = nRT is touched.
The second place it stops is anything with an exponent or a logarithm. Units do not cancel through a square root or a log. A pH is not a converted concentration; it is a transformed one, and its significant figures follow a different rule for the same reason.
The third is a relationship that only holds under conditions. Molar volume is the clean example: 22.41 L per mole is a perfectly good conversion factor, and it is a conversion factor for one mole of an ideal gas at 273.15 K and 1 atm and for nothing else. The formula pages carry those conditions beside every relationship, because a factor used outside its conditions cancels its units beautifully and still returns the wrong number.
The one habit worth taking from this
Write the unit on every number, including the ones in the factors, including on scratch paper. Most of what looks like a conceptual failure in a first-year course is a chain assembled in the wrong order, and the units say so a step or two before the answer does.
When a chain comes out wrong and you cannot see where, the question is which line the units stopped making sense on. Flasky answers that one directly: give it the problem and it writes the conversion out factor by factor, says why each factor was chosen and which way up, and leaves the cancellation visible. That is the form worth comparing your own line against, because the arithmetic was never the hard part.
Sources
- 1.6 Mathematical Treatment of Measurement Results, OpenStax Chemistry 2e
- 3.1 Formula Mass and the Mole Concept, OpenStax Chemistry 2e
- 3.3 Molarity, OpenStax Chemistry 2e
- 4.3 Reaction Stoichiometry, OpenStax Chemistry 2e
- Avogadro constant, CODATA 2022 value, NIST Physical Measurement Laboratory
- Abridged Standard Atomic Weights, IUPAC Commission on Isotopic Abundances and Atomic Weights