A “home” link is not protected by copyright in the first place.
The entire webpage design as a single work is.
Individual, original graphics used on the page are protected.
Any sufficiently long piece of text to be considered original and creative is protected.
Layout as a concept is not protected by copyright. You can’t copyright positioning of a navigation bar at the top of a page, or the names of a hierarchy of links. See Apple Computer, Inc. v. Microsoft Corporation. Menus and descriptive link names are obvious ideas, not original, creative expressions.
You still can’t copy any of the actual CSS or HTML from a web page in order to create a page with the same appearance.
What you need to do is to study their CSS to see how they created the effect you want and then clost their web page. You then write your own CSS to create the effect you want. While your CSS may include code that does things the same way as the other page does you have written it yourself rather than copying it from that other page. There may be pieces of it that look similar since it is creating a similar effect but none of the actual code was copied - only the idea for the particular effect.
In response to the OP’s question, there is no limit as to what you can copy design wise, though it’s highly unethical and frowned upon to do so. An exception to this rule would be if you were copying a trademarked brands website in order to impersonate them through the website design.
PS: For clarification IANAL but I know from reading past cases that website design copyright claims have been thrown out of court.
Basically it comes down to whether you actually copied the code or just happened to end up with near identical code because you are trying to do something that is almost identical. Even though the exact method of achieving a particular result may require particular CSS to be used there will still be differences in the exact order that the code appears and in naming of the classes that the effect uses.
While it may be difficult to prove that you copied code, rearranged a few properties and renamed a few classes rather than ending up with the same result yourself writing from scratch you’d have a hard time proving that you didn’t steal the code if the code is non-trivial and completely identical to the other site right down to the particular ordering of properties and weird naming used for the classes.
The ideal situation is where the similarities between your code and theirs are obviously there due to a similarity in the effect being created and is a lot less than the similarity between your code for creating that effect and your code for creating other effects. That way an independent person can see that your solution is written your way.
I still find it hard to believe that design on the web could be considered unique enough (as far as HTML and CSS goes - for the design) to qualify under copyright. The languages are limited enough that there is only a select number of combinations we can use and those combinations are further reduced because most designers fall upon conventions and patterns for specific implementations in order to maintain a websites usability. For example while a totally unique navigation method would be protected, who is going to use something so unique that it defies what the average user is going to understand, most common forms of navigation have already been well established and used widely which omits them from being protected. Another worry I have in that case is if you are claiming copyright on the design and it makes use of libraries or common conventions of code not protected (or owned by someone else) you are effectively claiming ownership of a design which has elements not owned by you and not fit for licensing, arguably that would degrade the value of the copyright on the design itself. Granted the US copyright office does state online works are protected (and that’s fine) but I haven’t seen any legal presidence to show that it would stand up in court that someone could be held liable for the copying of a design (rather than it’s contents or media), perhaps you have some examples?
Just changing the order of a few of the statements where order doesn’t matter, rearranging properties into a different order, and using slightly different id and class names would be sufficient to permit thousands of different ways to write the CSS to produce any one specific layout. So if two people were to create the exact same layout you’d expect there to be some differences in their CSS.
I would expect that ought to be enough to permit copyright to apply to a given arrangement although I agree with you that such things still haven’t been fully tested as to whether it actually does or not.