Scrolling: A horrible thing to do to humans who’re trying to read…

For hundreds of years, designers of books, magazines and newspapers have been evolving text to make it as readable as possible.

It wasn’t scientifically planned in the beginning. What worked, survived over the centuries, and what didn’t work died. Darwinian evolution.

Although it wasn’t scientifically planned, science can explain the evolution in terms of the human visual system.

A huge body of work on the readability of text was developed by two US university researchers in the years from 1929 until about the mid-1960s. They were Miles Tinker and his colleague Donald Paterson.

By as early as 1940, they’d carried out speed of reading and comprehension tests on more than 33,000 research subjects.

Although their findings were based on text in print, the same readability parameters apply equally to the screen – with a few adjustments.
Tinker and Paterson found reading only became comfortable when the text was at least 9 points high. (A point is the traditional printers’ measure, = 1/72 inch).

Readability improved at 10 point, got better still at 11 point, dropped off very slightly at 12, and fell off sharply beyond 13 point.

They also found that readers read most comfortably when columns of text were between 55 and 65 characters wide at those sizes.
Interestingly enough, the team I was then running at Microsoft funded a university research project which initially puzzled us until I realized that we humans will trade efficiency for comfort.
Mary Dyson and her team at the University of Reading in the UK found that reading was actually more efficient with wider columns, (~100 characters), but that people still preferred the narrower measure.
This surprising finding – and those of Tinker and Paterson, can be explained once you look at the human eye.


The picture above shows a cross-section of the human eye. Notice the spot at the back of the retina called the fovea. It’s a small dimple. In fact, it’s a bit distorted in this picture, because it has been deliberately made larger than actual size because otherwise you wouldn’t be able to see it.

It’s only 0.2mm in diameter, and makes a visual arc of 1.5 degrees with the lens of the eye.
It’s this dimension which dictates the size of test we can comfortably read.
Now look at this second picture:


This one shows the extraocular muscles which move our eyes from side-to-side and up-and-down. There are two other extraocular muscles which aren’t show; they are used to roll or rotate the eyes.

When we read text, we might think that we’re moving smoothly across the words, lines and pages. Nothing could be further from the truth…
It was a French oculist, Emil Javal, who found in 1906 that the eyes move in a series of rapid jerky movements.
Again, the size of the movements and their extent is determined by the size of the fovea, and controlled by the extraocular muscles.
Early research of this kind was pretty brutal. If you’re squeamish about eyes (like my Microsoft colleague and friend Greg Hitchcock), skip the rest of this paragraph. Javal attached small clay cups to the surface of his subjects’ eyes connected to threads to measure eye movements (some research projects used hooks instead!).
Greg: you can start reading again now 🙂
The findings of Javal and others have been confirmed by modern eye-tracking equipment which is a lot less invasive.
The eyes scan “targets” of between 5 and 7 characters at a time. They scan five such targets (Javal called them fixations) per second. They take only 20-25 milliseconds to jump jerkily from one target to the next (Javal called the jumps saccades).
It seems pretty obvious that the comfort v. efficiency finding that readers prefer columns of 55-65 characters is determined by the amount of tension and stretching required in the extraocular muscles which control side-to-side eye movement. If the tension and stretching gets uncomfortable enough, the reader has to turn her head – using a much larger set of muscles in the neck…
The earliest picture we humans learn to recognize is that of another human face. Here’s how the eyes behave when scanning it:


The black dots are the fixations, the lighter lines are the saccades.
In contrast, here’s how we move our eyes when reading text:



Saccade lengths vary: we can skip over common words like “and” , “A”, and “the”.
Notice the “reverse saccade” on line 2, where we may have to skip backwards if we either didn’t recognize or our brain didn’t properly parse the meaning from the pattern it recognized.


This simple analysis of human physiology explains why optimum line-lengths and leading evolved over the centuries to help us read.

If the distance between lines is too great, we have problems and are less efficient at finding the next line: the eyes have to travel too far, while the brain is holding the short-term parsing from the previous line so it can link the meaning to the next one.

At the same time, if the lines are too close together, two things happen: The ascending and descending strokes from adjacent lines can collide, or get close enough to confuse our pattern-recognition – or we end up reading the same lines twice, which confuses our brain as it tries to derive meaning from the patterns.

If you’re a proofreader, this phenomenon of reading lines twice has a name : it’s called doubling.

Now think about what happens when we read text in a scrolling window. We always end up reading some lines twice just to find out where we are.

Over the centuries, printers and typographers found (by evolutionary “experiment” – which we can call type mutations) that the optimum amount of spacing between lines of normal length was about 120% of the type size they were using.

They didn’t call it inter-linear spacing, of course. They called it leading, because it was created by placing thin pieces of lead between the lines of type.

When I started working in newspapers in Scotland in 1969, the newspaper I worked on was still set in hot metal. As a page editor, I had to be able to read type upside down and back-to-front in order to do the final copy-fitting on the “stone” – the heavy metal table (it used to be stone, hence the name) on which the frame holding all the type of the page was worked.

Now, we can’t blame the geeks who invented Web browsing in the 1990s for getting this wrong. It’s much easier to put text in a scrolling window than to paginate it – which of course you have to do if you aren’t scrolling. And it took print designers hundreds of years to develop the high standard of text readability we take for granted when we open a book today.

We’ve only really been reading text on screens since mass-market graphical user interfaces appeared with the Apple Macintosh in 1984 (I was there, with a Mac six weeks after they first shipped).

The Internet has really only been mainstream since about 1995.

In other words, we’re still in the early days of the Web. It’s easy to forget that.
And there’s lots that we can fix to make it better. In future postings, I’ll be exploring some of these in detail.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s