The Tailor's Unpicked Seam: On the Deliberate Release of an Unused Font Face
There’s a quiet satisfaction in a well-fitted garment, where every stitch serves a purpose. But what of the hidden seams, the ones tucked away in a lining, sewn with care but ultimately bearing no load? They are neat, they are precise, and they are entirely superfluous. In our stylesheets, we often craft such seams for our fonts, declaring faces with a tailor’s precision only to leave them folded away, unused. The browser, dutiful apprentice that it is, prepares them anyway, a small but measurable weight upon the page’s shoulders.
We are rightly taught the importance of `font-display: swap` to avoid invisible text, and of preloading critical typefaces. Yet, a more foundational oversight often persists: the declaration of font families that are never called upon. It is the `@font-face` rule that points to a weight or style we never actually use in our CSS. Perhaps it’s a bold italic from a library we only partially implemented, or a thin hairline weight from a abandoned design comp. They sit there, in the code, like unused spools of thread, and the browser, in its preparation, may fetch them, parse them, and hold them in readiness for a command that never comes.
The technique, then, is not one of addition but of subtraction. It is an act of editorial refinement. Before a stylesheet is considered complete, one must perform a simple but methodical audit. Is every single `@font-face` declaration actually referenced by a `font-family` property elsewhere in the CSS? This is not about visual inspection; it is a mechanical search. Use your code editor’s find function. Cross-reference each family name in your `@font-face` rules against every instance of `font-family` in your styles.
You will be surprised. A font named ‘Display-Light’ might be declared, while only ‘Display-Regular’ and ‘Display-Bold’ are ever used. That single unpicked seam, that Light face, represents an unnecessary network request, a slice of parsing time, and a fraction of memory. Multiply this by a few unused weights across a few typefaces, and the cumulative weight becomes a tangible drag on performance. Removing it is a pure gain, a reduction of effort with no aesthetic or functional trade-off. It is the quiet, uncelebrated work of optimization, the removal of something that should never have been there to begin with. It is the tailor, minutes before the final fitting, snipping away the basting threads that were only ever meant to hold things in place temporarily, leaving behind only what is essential to the garment's final form.
Notes & further reading
A few pages I came back to while writing this: