130

Researching specificity I stumbled upon this blog - http://www.htmldog.com/guides/cssadvanced/specificity/

It states that specificity is a point-scoring system for CSS. It tells us that elements are worth 1 point, classes are worth 10 points and IDs are worth 100 points. It also goes on top say that these points are totaled and the overall amount is that selector's specificity.

For example:

body = 1 point
body .wrapper = 11 points
body .wrapper #container = 111 points

So, using these points, I expect the following CSS and HTML to result in the text being blue: 

#a {
    color: red;
}

.a .b .c .d .e .f .g .h .i .j .k .l .m .n .o {
  color: blue;
}
<div class="a">
  <div class="b">
    <div class="c">
      <div class="d">
        <div class="e">
          <div class="f">
            <div class="g">
              <div class="h">
                <div class="i">
                  <div class="j">
                    <div class="k">
                      <div class="l">
                        <div class="m">
                          <div class="n">
                            <div class="o" id="a">
                              This should be blue.
                            </div>
                          </div>
                        </div>
                      </div>
                    </div>
                  </div>
                </div>
              </div>
            </div>
          </div>
        </div>
      </div>
    </div>
  </div>
</div>

Why is the text red when 15 classes would equal 150 points compared to 1 ID which equals 100 points?

Apparently the points aren’t just totaled; they’re concatenated. Read more about that here - http://www.stuffandnonsense.co.uk/archives/css_specificity_wars.html

Does that mean that the classes in our selector = 0,0,15,0 OR 0,1,5,0?

(my instincts tell me it’s the former, as we KNOW the ID selector’s specificity looks like this: 0,1,0,0)

Braiam
  • 4,345
  • 11
  • 47
  • 69
Sam
  • 4,217
  • 11
  • 38
  • 59
  • Here is something odd too : http://stackoverflow.com/questions/25565928/why-wouldnt-i-use-idid-to-add-specificity-to-a-css-selector – Armel Larcier Aug 29 '14 at 10:04

7 Answers7

142

Pekka's answer is practically correct, and probably the best way to think about the issue.

However, as many have already pointed out, the W3C CSS recommendation states that "Concatenating the three numbers a-b-c (in a number system with a large base) gives the specificity." So the geek in me just had to figure out just how large this base is.

It turns out that the "very large base" employed (at least by the 4 most commonly-used browsers*) to implement this standard algorithm is 256 or 28.

What this means is that a style specified with 0 ids and 256 class-names will over-ride a style specified with just 1 id. I tested this out with some fiddles:

So there is, effectively, a "point system," but it's not base 10. It's base 256. Here's how it works:

(28)2 or 65536, times the number of ids in the selector

  • (28)1 or 256, times the number of class-names in the selector
  • (28)0 or 1, times the number of tag-names in the selector

This isn't very practical for back-of-the-envelop exercises to communicate the concept.
That's probably why articles on the topic have been using base 10.

***** [Opera uses 216 (see karlcow’s comment). Some other selector engines use infinity — effectively no points system (see Simon Sapin’s comment).]

Update, July 2014:
As Blazemonger pointed out earlier in the year, webkit browsers (Chrome, Safari) now appear to use a higher base than 256. Perhaps 216, like Opera? IE and Firefox still use 256.

Update, March 2021:
Firefox no longer uses 256 as a base.

TylerH
  • 19,065
  • 49
  • 65
  • 86
Faust
  • 13,751
  • 9
  • 48
  • 107
  • 34
    Important: note that **the number 256 is not in the [spec](http://www.w3.org/TR/CSS2/cascade.html#cascade)**. Thus, this answer describes an (admittedly useful) implementation detail. – Matt Fenwick Aug 15 '12 at 18:09
  • 6
    Not only 256 is not in the spec as @MattFenwick said, but it also varies across implementation. It is apparently larger in Opera. In WeasyPrint and cssselect it is "infinity": I use [a tuple of integers](https://github.com/SimonSapin/cssselect/blob/v0.7.1/cssselect/parser.py#L88) instead of a single integer. – Simon Sapin Aug 16 '12 at 10:12
  • 3
    @Faust opera uses 16 bits instead of 8 – karlcow Aug 16 '12 at 10:21
  • 4
    This answer has a more practical description than Pekka's answer, to be honest. Basically what @Matt Fenwick says: what you're describing is a *practical implementation* of the spec. A flawed one at that, but not one that anything should be done about, be it by authors or implementers. – BoltClock Aug 17 '12 at 09:35
  • Great answer, for years I have thought I knew what specificity is, and I was wrong. But I don't get your last fiddle, as there isn't any id selector in it, just element selectors. – Fernando Jan 10 '14 at 10:28
  • 7
    256 appears to be insufficient in current versions of webkit (Chrome and Safari), further underscoring @MattFenwick's point. – Blazemonger Feb 12 '14 at 13:42
  • Can anyone comment/answer on this #id#id matter : http://stackoverflow.com/questions/25565928/why-wouldnt-i-use-idid-to-add-specificity-to-a-css-selector – Armel Larcier Aug 29 '14 at 10:03
  • 3
    Here to say that the provided fiddles no longer work as intended on Firefox, IE11 still does this though. – Baksteen Mar 09 '18 at 10:51
28

Good question.

I can't tell for sure - all the articles I manage to find avoid the example of multiple classes, e.g. here - but I assume that when it comes to comparing the specifity between a class selector and an ID, the class gets calculated with a value of 15 only, no matter how detailed it is.

That matches my experience in how specificity behaves.

However, there must be some stacking of classes because

.a .b .c .d .e .f .g .h .i .j .k .l .m .n .o

is more specific than

.o

the only explanation I have is that the specificity of stacked classes is calculated only against each other but not against IDs.

Update: I half-way get it now. It is not a points system, and the information about classes weighing 15 points is incorrect. It is a 4-part numbering system very well explained here.

The starting point is 4 figures:

style  id   class element
0,     0,   0,    0

According to the W3C explanation on specificity, the specificty values for the abovementioned rules are:

#a            0,1,0,0    = 100
classes       0,0,15,0   = ... see the comments

this is a numbering system with a very large (undefined?) base.

My understanding is that because the base is very large, no number in column 4 can beat a number > 0 in column 3, the same for column 2, column 1 .... Is this correct?

I'd be interested whether somebody with a better grasp at Math than me could explain th numbering system and how to convert it to decimal when the individual elements are larger than 9.

Stephen Watkins
  • 23,077
  • 12
  • 60
  • 97
Pekka
  • 418,526
  • 129
  • 929
  • 1,058
  • Thats because its between .o & .a .b etc So it would consider them collectively. But between an ID & a class e.g: .a and #a the ID will always overpower. It will (I assume) only count between classes when there is not a more overpowering attr. E.g class will go over element and id over class. – Sphvn May 11 '10 at 08:32
  • @Ozaki that is what I am assuming too, but it contradicts what the OP is saying about the points system. There must be more in play. I'd like to see the rules behind it. – Pekka May 11 '10 at 08:35
  • Just realized we've both come to the same conclusion. Excellent work! – Sam May 11 '10 at 08:48
  • @Sam thanks, I hope this is correct. Check out my update on the numbering system, it would explain why the id specifictiy behaves like it does. – Pekka May 11 '10 at 08:49
  • @Pekka That all makes sense except `96` where did that come from? :D – Sam May 11 '10 at 09:05
  • 5
    For the maths side, we can work in base 16 here (because none of the individual numbers exceeds 15). So 0,1,0,0 = 0100h = 256 0,0,15,0 = 00f0h = 240 256 > 240 so the id selector wins. – Matthew Wilson May 11 '10 at 09:08
  • 1
    Yes, you can think of specificity calculations as being done in a number system with a large base. I think the term "concatenating" (also used in the spec) is a much better description though. (Came here from answering a new question which turns out to be a dupe of this one, go figure...) – BoltClock Mar 02 '12 at 21:46
9

The current Selectors Level 4 Working Draft does a good job of describing Specificity in CSS:

Specificities are compared by comparing the three components in order: the specificity with a larger A value is more specific; if the two A values are tied, then the specificity with a larger B value is more specific; if the two B values are also tied, then the specificity with a larger c value is more specific; if all the values are tied, the two specifities are equal.

This means that the values A, B and C are completely independent of each other.

15 classes doesn't give your selector a specificity score of 150, it gives it a B value of 15. A single A value is enough to overpower this.

As a metaphor, imagine a family of 1 grand parent, 1 parent and 1 child. This could be represented as 1,1,1. If the parent has 15 children, that doesn't suddenly make them another parent (1,2,0). It means that the parent has an awful lot more responsibility than they had with just 1 child (1,1,15). ;)

The same documentation also goes on to say:

Due to storage limitations, implementations may have limitations on the size of A, B, or c. If so, values higher than the limit must be clamped to that limit, and not overflow.

This has been added to tackle the problem presented in Faust's answer, whereby CSS implementations back in 2012 allowed specificity values to overflow into each other.

Back in 2012, most browsers implemented a limitation of 255, but this limitation was allowed to overflow. 255 classes had an A,B,c specificity score of 0,255,0, but 256 classes overflowed and had an A,B,c score of 1,0,0. Suddenly our B value became our A value. The Selectors Level 4 documentation completely irradiates that problem by stating that the limit can never be allowed to overflow. With this implementation, both 255 and 256 classes would have the same A,B,c score of 0,255,0.

The problem given in Faust's answer has since been fixed in most modern browsers.

Community
  • 1
  • 1
James Donnelly
  • 117,312
  • 30
  • 193
  • 198
8

I am currently using the book CSS Mastery: Advanced Web Standards Solutions.

Chapter 1, page 16 says:

To calculate how specific a rule is, each type of selector is assigned a numeric value. The specificity of a rule is then calculated by adding up the value of each of its selectors. Unfortunately, specificity is not calculated in base 10 but a high, unspecified, base number. This is to ensure that a highly specific selector, such as an ID selector, is never overridden by lots of less specific selectors, such as type selectors.

(emphasis mine) and

The specificity of a selector is broken down into four constituent levels: a, b, c, and d.

  • if the style is an inline style, then a = 1
  • b = the total number of id selectors
  • c = the number of class, pseudo-class, and attribute selectors
  • d = the number of type selectors and pseudo-element selectors

It goes on to say that you can often do the calculation in base-10, but only if all columns have values less than 10.

In your examples, ids are not worth 100 points; each is worth [0, 1, 0, 0] points. Therefore, one id beats 15 classes because [0, 1, 0, 0] is greater than [0, 0, 15, 0] in a high-base number system.

Matt Fenwick
  • 44,546
  • 19
  • 115
  • 184
4

I am fond of comparison of Specificity ranking to Olympic Games medal table (gold first method — based first on the number of gold medals, then silver and then bronze).

It works also with your question (huge number of selectors in one specificity group). Specificity considered each group separately. In real world I've very rarely seen case with more than a dozen selectors).

There is also quite good specificity calculator available here. You can put your example (#a and .a .b .c .d .e .f .g .h .i .j .k .l .m .n .o) there and see the results.

Example of Rio 2016 Olympic Games medal table looks like enter image description here

rkarczmarczyk
  • 297
  • 3
  • 12
3

I don't believe that the blog's explanation is correct. The specification is here:

http://www.w3.org/TR/CSS2/cascade.html#specificity

"Points" from a class selector can't add up to be more important than an "id" selector. It just doesn't work like that.

Matthew Wilson
  • 3,723
  • 18
  • 13
  • Like I said in my answer. ^^ However it does make a difference if you have more of the same type (element, class, id). But that is pretty obvious if 5things say red and 3 say blue well Ima go red. – Sphvn May 11 '10 at 08:47
  • The wording around specificity probably hasn’t changed much between CSS2 and CSS2.1, but you really should be pointing to the CSS2.1 spec in future discussions as it completely supercedes CSS2 which was in general broken at the point of release. – Robin Whittleton Aug 16 '12 at 08:49
1

I would say that:

Element < Class < ID

I think they only stack into depending what you get if it is multiple of the same. So a Class will always overide the element and ID always over the Class but if it is down to which of 4 elements where 3 is to blue and 1 is to red it will be blue.

For Example:

.a .b .c .d .e .f .g .h .i .j .k .l
{
color: red;
}

 .m .n .o
{
color blue;
}

Should turn out red.

See Example http://jsfiddle.net/RWFWq/

"if 5things say red and 3 say blue well Ima go red"

Sphvn
  • 4,938
  • 7
  • 35
  • 55