What's the ratio?
I think this depends a lot on a many factors, but in general people will do both with research gradually taking precedence as time goes on.
In math Ph.D. programs, new students usually take classes for a grade as they prepare for qualifying exams. During this period they may not do any research. After passing the qualifying exams, this is the signal that they are ready to begin original research, but they may still take courses, usually ones that are more specialized and help prepare them for their thesis work; at some universities they will, however, not receive grades for courses any longer.
A lot of it depends on the country. In the US, you'll spend your first few years taking mostly classes for the masters, and even as a PhD student you might take classes. The research starts slow and builds up from masters to PhD level. In Europe, classes are rare in graduate programs (from what I've heard from friends studying in the UK and Germany, Italy). But they also make you finish the PhD in a few years, whereas you can take your time in the US.
It's also probably 'cause you already need a masters to start a phd in Europe, while most of the programs in the states are essentially combined ms/phd programs. The actual number of courses ranges, and tends to depend on the school as much as anything else. Usually the first two or three years end up being courses, but I've heard of schools requiring anywhere from 30 to 60 credits worth of grad work (and there are probably schools that require more than 60 or less than 30 credits-it's really incredibly variable.)
It is definitely something that varies from school to school, so I can only offer my own data point from personal experience. I am in a direct entry PhD program, and over the course of the 5 years, we must take 4 (3 credit) courses. So it is not all that much.
Edit: I should mention that my program is astronomy, not physics.
Another data point which is probably outdated: when I was a grad student at Michigan in the late 1970s / early 1980s, my "core classes" amounted to two per semester for about the first five semesters. Three semesters of quantum, two of E&M, one each of classical mechanics and thermo/stat mech, and two math courses. After that, I think I had to take at least one class every two years (four semesters) until I finished my degree.
A "full" course load was four classes. I was technically a half-time student, because I had a half-time assistantship which is what I was paid for (teaching during the first two years, research afterwards). After I started my dissertation research, I registered for "dissertation research" as if it were a class, for enough hours as necessary to maintain my half-time student status.
This was for direct entry from a bachelor's degree, which is usual for Ph.D. physics programs in the U.S.
wow! classes every term? What kind of class could you take as a 5th year phd that you would waste time on not researching?
At all the schools ive seen, its basically year one being core classes (quantum, stat mech, e&m, etc) and year two being specializing (QFT class, comology class, solid-state, etc)).
When I was a grad student, we had to meet a core requirement, and we had to take a certain minimum number of courses.
Core requirement: two semesters of quantum mechanics; three from two semesters of statistical mechanics, two semesters of quantum field theory, one semester of elementary particle physics; two one-half semester advanced special topics courses; four graduate seminar presentations (on anything except one's research topic).
I can't remember the minimum number of courses required. I took three graduate courses in pure mathematics in groups, Lie groups, and representation theory (offered by the math department and approved by the physics department), and I took all five of the three of five part of the core requirement, so I took five courses beyond the core requirement. This might have been a little more than the minimum requirement.
Grad students who had not had Goldstein/Jackson-level courses had classical mechanics and electromagnetism added to the core requirement.
My biggest disappointment in terms of courses as a grad student: There was a grad course in general relativity which was on the books, but which was was not offered during my entire time as a grad student. It was offered shortly after I finished grad school.
First two years it's mostly classes. After the first two, you are "all but dissertation" so it's mostly research. Also, you will be spending pretty large amounts of your time teach classes and grading papers, at least for the first few years.
Separate names with a comma.