That is perfectly normal. Since columns don't run dry, the first elution step is a mix of elution buffer and wash buffer and therefore can't elute so much protein. The second and third one have a higher imidazole concentrations and elute better. You can try to control this process by using a step-wise elution buffer with increasing imidazole concentrations, that way you can identify the optimal fraction concerning purity.

As a rule of thumb, when a protein BEGINS eluting - that is, the buffer conditions are right for complete unbinding from the matrix - the first hint of protein will occur ONLY after the dead volume.

"Dead Volume" = Assume a protein does not bind the column at all, or is completely begun unbinding the matrix. It will take time for it to travel from the top of the column to the bottom. The volume of buffer required for this is the column "dead volume". Any protein eluting before the dead volume would indicate the column is faulty. Thus, when you load your CRUDE extract FOR THE FIRST TIME, the first fraction of the flow through that contains protein will determine the dead volume. Again, rule of thumb dead volume is about 30-35% of bed volume.

"Elution Volume" = Volume required to COMPLETELY elute protein and is usually = 2 x bed volume, ASSUMING the protein is completely eluting with hat buffer. So it is the sum of dead volume + volume of fractions that contain protein. So your first fraction, at the time of adding elution buffer should NOT contain any protein, because it is the dead volume. If your fraction volume is > dead volume, but < elution volume the first fraction will have less protein than the next fraction.

Let us assume you collected 3 fractions of 1 ml each for the above expt you mention. Now if you had collected 6 fractions of 0.5 mL each, then your protein will be "maximum" in between fractions 3 & 4. It is a VOLUME thing and not a fraction number thing. The more fractions you collect for the same total elution volume means you have to collect less and less volume per fraction.

There is an advantage is stretching out fractions with smaller volume size. This allows, for finer "separation" FRACTION wise (or volume wise) between two proteins that elute in very close. Thus, if you elute for 1 ml, if protein A elutes in 0.5 mL and protein B immediately following it, then using two 0.5 ml fractions gives you a better chance of separating A from B in different fractions. If you collected all 1 ml in same fraction, A will be contaminated with B.

Think this through, it is not that difficult !! :-)