Choosing a Ratio for Your Type Scale

The type-scale generator on this site offers eight named ratio presets, borrowed from musical-interval names, and it's tempting to just pick whichever name sounds most appealing. The names don't tell you much on their own — what matters is what each ratio actually produces once you run real numbers through it, and how that interacts with the base size you start from.

All eight presets, for reference

In order from gentlest to most dramatic: minor second (1.067), major second (1.125), minor third (1.2), major third (1.25), perfect fourth (1.333), augmented fourth (1.414), perfect fifth (1.5), and the golden ratio (1.618). Each is simply a multiplier applied repeatedly — step n's size equals the base size times the ratio raised to the power of n — so the only real decision is how much size difference you want between adjacent steps in your hierarchy. A separate guide on this site works through a comparison of minor second, major third, and perfect fifth in detail; this one covers three different ratios from the same list, so between the two articles you can see the full spread.

Three more ratios, from the same 16px base

Generating a minor third (1.2), a perfect fourth (1.333), and the golden ratio (1.618), all from a 16px base with two steps down and four steps up, makes the gap between "gentle" and "dramatic" concrete. The minor third reaches 33.18px four steps up — a little over double the base. The perfect fourth reaches 50.52px over the same four steps — roughly three times the base, and noticeably more headline-sized. The golden ratio reaches 109.66px over those same four steps — nearly seven times the base size, a genuinely dramatic jump that would suit a single commanding poster headline far better than a UI with several closely related heading levels. Looking at the bottom end tells the same story in reverse: two steps down, the minor third only shrinks to 11.11px, while the golden ratio shrinks all the way to 6.11px — a size too small to be legible as running text, which is a useful signal that the golden ratio's steps-down direction is rarely worth using much past one step for practical type.

The same ratio, two different bases

Ratio and base size are independent choices, and it's worth seeing how the same ratio behaves differently depending on where it starts. Take the perfect fourth (1.333) from a 14px base instead of 16px: four steps up lands at 44.2px, versus 50.52px from the 16px base — roughly a 6px difference in the resulting headline size, purely from a 2px difference in where the scale started. From an 18px base, the same four steps reach 56.83px. None of these are wrong; a smaller base is a reasonable choice for a screen-first design where body text needs to stay compact, and a larger base suits a print piece meant to be read at a greater distance. The point is that base size and ratio aren't two independent knobs you can tune in isolation and expect predictable results — changing the base reshapes every size the ratio generates, proportionally, all the way up and down the scale.

How many of the generated steps are actually usable

Generating a wide range of steps doesn't mean every step is a size you'd actually use. Take a minor third (1.2) scale from a 16px base with three steps down and six steps up: the full range runs from 9.26px at the bottom to 47.78px at the top. If you set a rough practical floor of about 14px (a widely used comfort threshold for sustained reading) and a rough ceiling of 40px (past which a size reads as clearly a headline rather than an unusually large paragraph), only about seven of this scale's ten generated steps — from roughly 13.33px through 39.81px — fall inside that practical window. The two steps below it (9.26px, 11.11px) are too small for comfortable body text, and the top step (47.78px) is well into headline territory where a different leading and tracking treatment applies anyway. This is worth checking on any scale with a wide step range: generating ten sizes doesn't mean you have ten usable options, and it's worth identifying which generated steps are genuinely candidates for a role in your hierarchy before you start assigning them.

A practical way to choose, rather than guess

Rather than picking a ratio by its name or its reputation, it's more useful to generate two or three candidates at your actual intended base size and step count, and look at where the extremes land. If your tightest useful step and your largest useful step both fall within a workable range for your medium — comfortable body text at the small end, a headline that fits your layout at the large end — the ratio is doing its job. If the top step blows past what your layout can hold, or the bottom step drops below comfortable reading size before you've used all your steps down, that's a concrete, checkable signal to try a gentler or steeper ratio rather than a vague aesthetic judgment about which name "feels right."

Where line height and tracking come back in

A generated scale only gives you sizes; it doesn't tell you the line height or tracking each size should carry, which are separate calculations covered in their own guides on this site. Once you've settled on a ratio and base you're happy with, the type-setting reference runs a full scale through those follow-on calculations in one table, so you can see size, recommended line height, tracking, and baseline-grid placement together for every step, rather than recalculating each one separately as you assign roles to the sizes you've generated.

A worked scenario: fitting a fixed headline space

Say a poster layout has a fixed maximum headline height of 60px — anything larger physically doesn't fit the space reserved for it. Generating a perfect fourth (1.333) scale from a 16px base, four steps up lands at 50.52px — comfortably inside the 60px limit, using most of the available room in just four steps from the base. Generating a minor third (1.2) scale from the same base, it takes seven steps up to reach a comparable size (57.33px), and the eighth step (68.8px) overshoots the limit entirely. Neither ratio is wrong, but they solve the "fit this space" problem very differently: the perfect fourth reaches a usable headline size in a small number of steps, leaving fewer, more widely separated levels available for the rest of the hierarchy; the minor third needs many more steps to reach the same neighbourhood, which gives you more finely graded intermediate sizes along the way, at the cost of needing to climb further to get there. Checking a specific space constraint like this against a couple of candidate ratios is a much more concrete way to choose than judging the ratios in the abstract.

Odd step counts and a centred hierarchy

One small, easy detail: if you want your base size to sit visually in the middle of your generated range rather than at one end of it, choose equal steps down and steps up (say, three and three, or four and four) rather than an asymmetric split. An asymmetric scale isn't a mistake — the modular-scale guide on this site covers deliberately weighting a scale toward one direction when a project genuinely needs more range at one end — but if your goal is simply "a base size with a roughly symmetric ladder of options above and below it," matching the step counts is the straightforward way to get that, and it's worth double-checking your step counts actually match if a centred base is what you're after, since it's an easy parameter to leave lopsided by accident.

Treating "golden ratio" as just another number

The golden ratio (1.618) tends to arrive with more mystique attached than the other seven presets, and it's worth setting that reputation aside and treating it as exactly what it is here: a multiplier, functionally no different from 1.5 or 1.333 in how the generator uses it, just a larger one. It's not a guarantee of a more harmonious result, and a scale that's too dramatic for a given project doesn't become correct because the ratio it used has a well-known name. Judge it the same way you'd judge any other candidate — by generating it at your actual base size and step count and checking whether the resulting sizes fit your layout — rather than reaching for it because of its reputation outside of type design.

A short process, repeated for every project

Put together, the process this article has worked through is short enough to run on every new piece rather than being a one-time decision you make once and reuse forever: pick a base size suited to the medium, generate two or three ratio candidates at your actual step count, check where the extremes land against your real constraints — a legibility floor at the small end, a physical space limit at the large end — and pick whichever candidate uses that available range most sensibly. It takes a few minutes with the generator and produces a scale you can explain, in specific numbers, rather than one you can only defend by pointing at a ratio's name.