We ran some queries for the ~300k websites that are using only 9 elements per page and here are the numbers we’ve got for the top 15:

meta 2,056,149
script 687,084
link 369,959
div 346,548
head 253,189
html 252,865
title 252,751
body 251,831
p 131,922
h1 116,473
style 112,121
a 72,571
li 53,858
br 15,975
frame 14,414

The above are the total number of HTML elements found within ~300k pages.

Neopets may have skewed the averages a little. They haven’t done a redesign in many years, and every single page of theirs is formatted with several nested tables. Every. Single. Page. Has. Several. Nested. Tables. It makes me dry-heave just thinking about it, and it’s where I first learned to write HTML and CSS as a kid. My code is written as bare-bones, literal, and minimalist as possible in silent rebellion to this day.

Joan, hard to tell but not impossible :)

So the 15 most common uses of the <i> tag made up about 5.5 million of 26.2 million <i> tags.

How many of the <i> tags didn’t have a class and had something between the opening and closing tag? That is, how many were likely being used for something other than an icon font element?

Within the total number of inline elements (936,760,353), 2.8% (26,229,289) of them are ielements.

So, I dug a bit deep into the i element stats only and here is the top 15 for <i class="*"></i>:

fa fa-angle-right – 696,960
fa fa-twitter – 561,335
fa fa-clock-o – 543,666
fa fa-facebook – 498,275
fa fa-angle-down – 455,119
ddc-icon ddc-icon-chevron-right – 448,404
fa fa-user – 439,849
fa fa-star – 409,194
fa fa-search – 397,014
fa fa-google-plus – 307,473
fa fa-shopping-cart – 307,700
ddc-icon ddc-icon-arrow2-right – 296,082
fa fa-bars – 254,183
fa fa-heart – 241,411
fa fa-envelope – 244,036

Font Awesome FTW!

Within the total number of inline elements (936,760,353), 2.8% (26,229,289) of them are ielements.

I dug a bit deep into the i element stats only and here is the top 15 for <i class="*"></i>:

fa fa-angle-right – 696,960
fa fa-twitter – 561,335
fa fa-clock-o – 543,666
fa fa-facebook – 498,275
fa fa-angle-down – 455,119
ddc-icon ddc-icon-chevron-right – 448,404
fa fa-user – 439,849
fa fa-star – 409,194
fa fa-search – 397,014
fa fa-google-plus – 307,473
fa fa-shopping-cart – 307,700
ddc-icon ddc-icon-arrow2-right – 296,082
fa fa-bars – 254,183
fa fa-heart – 241,411
fa fa-envelope – 244,036

Font Awesome FTW!

So yeah, 99% English

Of the 8,021,323 pages that we were able to parse, 5,368,133 use the lang attribute on the html element. That’s about 70%!

Check out below the breakdown for lang attribute values:
– en-US 2,688,150
– en 2,104,991
– en-gb 267,844
– en-GB 120,406
– en-us 54,480
– de 40,250
– fr 26,156
– en-AU 24,133
– es-ES 21,561
– fr-FR 20,162

I just tested with a simple table, and Chrome, Safari, and Firefox all insert tbody elements automatically.

It absolutely classes as surprising. There being a valid use case for tables certainly doesn’t equate to there being a need for tabular data on a third of all websites.

I can barely even think of examples of what an average website would show in a table.

What’s also surprising is that it seems like there are websites out there that use <table> but not <td>

Stephen,

That’s very interesting.

Right now I’m only thinking about checking out the “divitis” scenario for all the ~300k websites that are using those 9 different elements only.

Using only 9 types of tags is not necessarily “short and clean” and could just as easily be long and dirty.
I imagine two likely cases for a curious spike at exactly N tags (here N==9) and am interested to know if Catalin notices if either is the case.

a) divitis
— I see many sites with deep nesting of <div>s as wrappers of wrappers of wrappers, often as part of building a site based on a framework system. You can often tell which framework too, based on the classes on the divs.

b) Automation
— If you have E-Z Site Builder v3 churning out pages according to a template, many similar pages & sites would inflate the data at the N point. Since automation generally can’t infer semantics things tend to turn into generic divs.