Adventures in Space: Spacing

February 04, 2016 by Yves Peters

William of Wykeham, founder of Winchester College in 1382, made the motto of the college “Manners maketh man”. Similarly one could say that “Spaces maketh font”. Indeed, the white bits of nothing between the characters are just as important as the black. Without proper spacing a typeface is nothing but a collection of glyphs, not a font. In this upcoming series of posts, I will look at spacing, kerning, tracking, and the different types of spaces.

The top line shows Bruno Mello’s Objektiv with the proper built-in spacing, as defined by its designer. In the middle line all spacing was removed. Setting the side bearings to zero makes the bounding box exactly as wide as the glyph. Certain character pairs crash into each other, while others still have some space between them thanks to their structure. Note how there is still a sliver of white left and right of the ‘i’ because the tittle is wider than the stem. In the bottom line the space between the characters, measured from one extreme point to the next, was made mathematically identical. This creates very uneven spacing.

The actual drawing the hundreds of glyphs required for a performant contemporary font may seem like an enormous amount of work, yet it is only half the effort involved in the development of a typeface. It is the quality of the spacing (and to a lesser extent kerning), the careful arrangement of space around the characters, that distinguishes a good typeface from a poor one. Without proper spacing, a typeface is nothing more than a barely usable collection of black shapes alternatively drifting apart or colliding into each other.

Drawing is Spacing

The white between glyphs (here in yellow) in César Puertas’ Buendia is balanced against the white inside the glyphs (here in red). The mathematical distance between consecutive glyphs varies quite a bit – compare for example the distance between the extreme points of the ‘e’ and ‘s’ with the outstroke of the ‘d’ that almost touches the bottom left serif of the ‘h’. However the optical distance has been optimised to look perfectly balanced.

Reading involves deciphering the interaction between the black of the letter forms and the white surrounding the black. There are two types of space – the white inside the letters (except of course in letters like the capital ‘I’, the lowercase ‘l’ or the number ‘1’), and the white outside them. When drawing character shapes a typeface designer has to continuously balance those two types of space in order to create an even grey value of the text; the amount of white inside a glyph influences how much white needs to be outside the glyph. This ‘grey value’ determines how pleasant and effortless the reading experience will be. This is why it is vital to space the glyphs as you are drawing them when designing a typeface. Drawing and spacing are intrinsically connected and need to happen simultaneously.

The different character shapes create different shapes of white inside them. Green shows spaces that are entirely enclosed by the glyph. Blue shows spaces that connect with the space outside the character. For the purpose of this article we consider spaces that have an opening at the bottom or the top also as enclosed spaces.

The design of a typeface relies on the tension between difference and similarity. The character shapes need to be similar to achieve a harmonious appearance, yet different enough from each other to be easily identifiable. When we examine the characters in a typeface, we see these different shapes create different types of space – for example the white can be entirely inside the shape, like the counter in the ‘o’, or can be partly inside and partly outside the shape, like the white between the bowl and the loop in a double-storey ‘g’. All the glyphs also have white outside the shape: left and right, above and underneath. For the purpose of this article we will restrict ourselves to the white left and right.

Controlling Halves

Similar to the characters in its physical counterpart [Real](/content/printing-with-real-wood-type) that sit on pieces of wood, every glyph in the digital [FF Real](https://www.fontshop.com/content/interview-with-erik-spiekermann-and-ralph-olivier-du-carrois-1) is positioned in a virtual rectangular area called the ‘bounding box’. The distance from the left or rightmost extremity of a glyph to the edge of the bounding box (here in red) is called the side bearing (here in darker yellow). While the height of the bounding boxes is constant in a typeface, the width of each one is defined by the width of the glyph plus its side bearings. — Similar to the characters in its physical counterpart Real that sit on pieces of wood, every glyph in the digital FF Real is positioned in a virtual rectangular area called the ‘bounding box’. The distance from the left or rightmost extremity of a glyph to the edge of the bounding box (here in red) is called the side bearing (here in darker yellow). While the height of the bounding boxes is constant in a typeface, the width of each one is defined by the width of the glyph plus its side bearings.

The basic challenge the typeface designer faces is that she has control over only half of the space to the left and half of the space to the right at a time. Because the characters, not the spaces are the basic units in a typeface, the spaces between the characters need to be cut in two so that individual characters can be arranged to set text. The designer needs to decide where the left half of the space ends and stick it to the right side of the character on the left. The other half of the space belongs to the left side of the character on the right. You can easily visualise this when you look at metal type. Today’s bounding box around each character is a digital representation of the physical piece of lead the character used to sit upon. The distance from the left or rightmost extremity of a glyph to the edge of the space is called the side bearing.

For any space next to a character, only half of it is defined by the character itself and the other half belongs to the preceding or the following character. As you can see in this example showing a number of possibilities for the ‘o’, this creates whites of wildly varying shapes (here in yellow). For the purpose of this visual the kerning was disabled.

Thus the space between any given character and the preceding one is actually created by sticking together two half spaces – the space at the left side of this character with the space to the right side of the preceding character. The opposite applies to the space between the character and the one following it. This is why the spaces left and right of any given character are ever-changing, as they depend on what other glyphs precede and follow that character. The only way to tackle this problem is to constantly check the spacing in test strings of different sequences of characters during the drawing process. Most typeface designers start with a round shape like the ‘o’ and a straight-sided shape like the ‘n’ and expand from there. You cannot first draw all the glyphs and then come back to define the side bearings afterwards. This is the reason why drawing an alphabet in Adobe Illustrator before importing it in a font editor might not be a good idea.

Making Sense of Spaces

Even though the humanist letter forms in [Toshi Omagari](/designers/toshi-omagari)’s Venetian [Marco](/content/best-fonts-of-2015) are far more intricate and ‘irregular’ than the comparatively simpler ones in sans serifs, each vertical row shows how the marked spaces are of a similar shape. This phenomenon is very helpful in spacing glyphs consistently throughout a typeface. — Even though the humanist letter forms in Toshi Omagari’s Venetian Marco are far more intricate and ‘irregular’ than the comparatively simpler ones in sans serifs, each vertical row shows how the marked spaces are of a similar shape. This phenomenon is very helpful in spacing glyphs consistently throughout a typeface.

Character spaces can be divided into a number of basic categories according to their shapes: rectangular ones next to characters with vertical sides, curved ones next to round characters, triangular ones, and then a few exceptions for unique characters like ‘a’, ‘g’, ‘s’, ‘x’, etc. This helps type designers introduce consistent spacing in characters with similar shapes – for example the left bearing of an ‘e’ will be similar to the ‘c’, ‘d’, ‘o’, and ‘q’, and the right bearing of an ‘h’ will be similar to the ‘m’ and ‘n’. Unfortunately this system will get you but so far, and it takes the eye of a skilled type designer to space all the glyphs in a typeface well. The quality of the spacing often gives away the level of proficiency of the type designer.

Even with the most careful spacing you cannot provide a solution for every single character pair. To illustrate this, kerning was disabled in the top line. While the ‘Th’ pair is fine, the lack of ascender in the ‘Ta’ pair creates an unsightly gap between the two characters. Activating the kerning in the bottom line makes the bounding boxes overlap, allowing the ‘a’ to slide underneath the overhanging arm of the ‘T’.

Yet even the best type designer cannot find a perfect spacing solution for every single character combination. As a last resort she will have to kern specific glyph pairs to correct the unavoidable shortcomings in the spacing. This is the topic of the next article in this series.

Top line: Pieter van Rosmalen’s monospaced Nitti. Bottom line: Its proportional sibling Nitti Grotesk.

Due to their very nature different rules apply to monospaced (also called fixed-width) fonts. In proportional typefaces the bounding boxes are of varying widths: they are defined by the width of the glyph plus the side-bearings. As their name implies the bounding boxes in monospace fonts all have the same fixed width, and no kerning is needed. The character shapes need to adapt within this fixed space to distribute the inner and outer white space as best as possible. This produces a much more uneven grey value.

Foreign Spaces

Because I don’t really feel qualified to discuss spacing in non-Latin scripts, Latin-based alphabet systems (so Cyrillic and Greek as well) are the focus of this article series. Generally speaking most scripts tackle spacing in similar ways, adapted to their specific structure and rules.