Which font and line length is most readable on the screen?

Everyone of us have many times heard or said statements like:

  • Sanserif fonts in small sizes are easier to read on the screen then serif fonts.
  • Verdana is the best existing web font.
  • Arial is good in putting much info into a small space.
  • The best line length is 40-60 characters.

I tried to find if there was any scientific source behind these common statements. When I could not find it, I asked two students (Julia Barrantes and Elahe Jalili Baleh) at my university to make their own experiments.

In their experiments, 20 people, all with some computer experience, were inititally shown some texts in Times New Roman, Verdana and Arial, in different sizes, and asked which was the smallest easily readable size. In this test, the median agreement among the respondens was:

  • 8 pt (=10px) for Verdana,
  • 9 pt (=12 px) for Arial and
  • 10 pt (=13 px) for Times New Roman,

assuming 96 pixels/inch.

Interesting here is that the same text in Verdana 10 px (=12 pt) and in Times New Roman 10 pt(=13px) requires exactly the same space in pixels on the screen, while Arial 9 pt (=12 px) requires more space on the screen, so the statement that Arial is best in putting much readable text into a small screen space seems to be wrong!!

After that, the same 20 people read a number of very similar texts in different fonts, line lengthes and line spacing. Their main result were:

Comparison Result Statistical significans
Comparing 8pt Verdana with 10 pt Times New Roman, 40 characters/line, 120 % line distance. Times New Roman gives 7.45 % faster reading. t=0.026, i.e. a probability of 2.6 % that the conclusion is in error.
Comparing 8 pt Verdana with 9 pt Arial, 40 characters/line, 100 % line distance. Arial gives 3.45 % faster reading. t=0,29, i.e. not significantly significant difference from the hypothesis that both fonts give the same reading time. But there is certainly no statistical significance for the reverse conclusion often heard.
Comparing 40 and 80 characters/line, 120% line spacing, font 8 pt Verdana. 40 characters/line gives 8.5 % faster reading lthan 80 characters/line. t = 0.027, i.e. a probability of 2,7 % that the conclusion is in error.
Comparing 40 and 80 characters/line, 120 % line spacing, font Times New Roman. 80 characters/line gives 2.24 % faster reading! t=0.842, i.e not statistically significant that 80 Characters per line gives faster reading than 40 characters line. But there is certainly no statistical significance for the reverse conclusion often heard.
Comparing 40 and 80 characters/line with 120 % line spacing and 9pt Arial font. 80 characters/line gives 1.05 % faster reading. t=0.847, i.e. not statistically significant difference from equal reading speed at both line lengthes.

Conclusion

Verdana may not be the best web font. At least in our test, Times New Roman 10 pt gave faster reading than Verdana 8pt. (Verdana 8pt is comparable to Times New Roman 10 pt in the actual character sizes and number of pixels used for the same text.)

40 characters/line can be read faster than 80 characters/line for Verdana. But for Times New Roman, and with 120 % line spacing, Times New Roman does not give slower reading with 80 characters/line than with 40 characters/line, it may even give faster reading with 80 characters/line.

A demo page can be found at http://www.dsv.su.se/jpalme/internet-course/minimum-sizes-of-fonts-v4.html.

If you want to repeat this experiment, we found that the compared texts must be very similar, to reduce error introduced by difference between the test texts. We used a simple fairy-tale story, and made the pages different by just replacing some words in the text, like "mother" to "father", etc. In this way, the syntactic structure was the same for all the test texts.

Note: The tests were made on Swedish language texts read by Swedes. Swedish has a rather similar structure as English, so I do not think the language is important for the results. However, people may read text faster in fonts they are accustomed to, and Serif fonts may be more commonly used in Sweden, while sansserif fonts may be more common in America than in Sweden, and this may mean that the results would be different for American people reading English texts.

Author

Written by Jacob Palme, based on a study (only available in Swedish) made by Julia Barrantes and Elahe Jalili Baleh.

Location of this page

http://dsv.su.se/jpalme/internet-course/font-report.html