I always heard that in web design one should not prefer pt as a size unit, because the browser/OS does not neccessarily know the correct dpi. Result: final font sizes are not equal on different computers (Mac and Windows have different default dpi, for example).

But isn't this true for the "prefered" px as well? If I buy a FullHD monitor with 20" diameter and a FullHD monitor with 40" diameter, the font sizes will differ by a factor of two, because the physical pixel sizes are twice the size on the bigger monitor compared to the smaller one. So, for px I have the same effect, because the browser does not know the device size.

Conclusion: If I take a ruler and measure a pt based font on two different devices I can get different results (unknown real dpi size). But for sure I can get different results for a px based font size as well just because the same-resolution-device is bigger/smaller (unknown real physical pixel size).

So, why is the "unknown device parameter argument" used for pt but not for px?

Bonus question: why do devices not deliver their physical size information via its device driver to the OS, anyways? A device should know how big it is and how much physical pixels it has, because someone build it with a defined size and pixel density and could just store this in the firmware.

  • As for physical size, there is a problem with projectors. The size changes based on how close the projector is to the wall. Even if there were a built in default size, it wouldn't be appropriate to use for web page development. max-width: 12in would be ridiculously small projected on a wall while being pretty big for most monitors. Apr 2, 2015 at 13:11
  • @StephenOstermiller Shouldn't a projector (approximately) being able to calculate the distance to the wall on the fly by using the lense parameters (e.g. focal length), assuming the user adjusts these parameters correctly to make the image appear sharp?
    – Foo Bar
    Apr 2, 2015 at 13:24
  • First, that technology would have to be built into technology. While it sounds possible, it might make projectors more expensive. I don't know that any projectors currently do it. Then it would have to be able to communicate that information back to the computer. I'm not sure that DVI or HDMI would even support that. Then the computer would have to make that information available to the software. Even if all that happened, I don't think it would be useful information. Generally you use a projector because you want them blown up so everybody can see. Apr 2, 2015 at 13:28
  • Actually, this technology exists since 1994: en.wikipedia.org/wiki/Display_Data_Channel - questions is, why it is not used widely. Anyways, this was just a bonus question, the main question assumes that this data exchange is not possible or not used.
    – Foo Bar
    Apr 2, 2015 at 13:34
  • Are you talking device pixels or CSS pixels? They are not the same. There IS a formula for pixels as defined by the W3C and applies to the web. But it's too much for me to go into here and I think this may be off topic.
    – Rob
    Apr 2, 2015 at 23:13

1 Answer 1


Good question. I hope i got this right, seems like the point (DPI) is more volatile to digital display environment whereas a pixel is more static.

A pt is much older and comes from the print/lithography era. Classically it is 1/72 of an inch. So in that regard an inch is an inch is an inch...except when it goes digital. In order for a device to deduce what an inch is, it must know its own pixel size, pixel density, resolution, physical size, relativity, etc. These are 4-5 factors of "device parameters" that affect the "real world parameter" of an baseline inch in a conditional(ish) manner. A point could equal 20 pixels or a point could equal 200 pixels, depending on device, but it is always equiv to 1/72 of an inch (which is why you often see 72 DPI for digi displays).

A px is more simple: its a pixel without regard for density, inches, etc. This means that there are no "device parameters" to affect the conditional calculation of said px. Retina seems like it would be different, but its not, its still just a pixel in a normal-density alpha transparency layer. The retina layer is separated thang that doesnt really affect the proportions of px itself. So 20 pixels could equal 1 point (dot) or 200 pixels could equal 1 point (dot), but 1 pixel is always equivalent to 1 pixel.

As far as the bonus question goes, I think it comes down to: "Life is unfair, get over it. Just kidding. We werent thinking too hard about the future and didnt expect multi-device resolution to become such a critical thing, therefore didnt worry about coding up good identifiers for developers to use. We have started to get better at this and are now sharing some better data from screens"

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.