Do I have it right about density gradient centrifugation of lipoprotein?

Context: I am a statistician analysing data coming from a DGUC of lipoproteins and I would like to understand exactly what I'm looking at. To wit, the data I am analysing comes in the form of "curves" (1 curve per individual) expressing cholesterol density (mg/dL) in terms of "fraction". The graph below represents 4 such curves.

From my understanding, the idea of density gradient ultracentrifugation is this:

1° In a test tube, prepare a solution of water + "something that disolves in water (say sucrose)" whose density (mg/ml) varies continuously from lowest density (bottom) to highest density (top). This can be achieved by simultaneously adding to the test tube a solution of water to a solution of high concentration water+sucrose with the water being added at a constant rate while the  water+sucrose is added at an increading rate. In this step we have to make sure that the range of densities covered includes the densities of the objects we are trying to isolate.

2° Add the objects we are trying to isolate on top, attach the tube to a wheel along its radius and spin the wheel in order to induce on each particle in the tube a force parallel to the lenght of the tube and oriented towards the top of the tube. Each type of object that we are trying to isolate has a characteristic density (its total mass divided by its volume). This will cause each object to migrate to the layer of liquid of density equal to its own. The higher up in the tube, the greater the density. Once the migration process has stabilized, stop the wheel.

3° Use a pipette (or some other less disturbing mean) to separate the content of the test tube into a predetermined number of "fractions" of equal volume V.

4° And here I'm really not sure what action is taken to produce the graph that we see. My best guess is that the lipoproteins inside each fractions is somehow weighted and if W_i is the weight in fraction i then W_i/V is computed in units of mg/dL and it is this number which makes up the Y coordinate of the points in the graphs. In particular, if this is indeed how step 4 goes down, it would mean that the actual number on the Y axis are really meaningless since, for instance, the numbers would be twice as big had we added twice as many objects in step 2.

Do I have the steps right? What about 4°? If I'm right about 4°, how are the objects weighted?

Post Merge: A month ago