Suppose you are using an arrangement as shown at the cited site.  Due to symmetry, at the center between permanent magnets there should be no perpendicular component of magnetic induction, only the one along the line connecting centers of permanent magnets.  This may change when you move your sensor along the said line, but should still hold in the symmetry plane.

Thanks for your assistance. That means- though the attracting lines between the two magnets will not be fully perpendicular on any sample in the middle, but the flux (magnetic lines of force) of both the magnets will eventually add up in attracting direction (As in, N faces S pole). Have I got it correctly? ~~ Still not sure, how much the iron/steel path is going to to contribute into the flux density, or do they really contribute at all?    

Think about a bar magnet.  The strongest field it creates is, of course, near its poles.  On the other hand, the B (or H) lines are always closed, hence the weak field is present near magnet's sides.  Moreover, this field is directed oppositely to the one inside the magnet.  You may think of this magnet as of the iron yoke in your apparatus.  Yoke's field will then slightly decrease the field between permanent magnets.  I bet its share will be neglectful.   I wouldn't worry about that.  It's quite likely that room temperature variations may influence the field between permanent magnets even more.  To make your working area filled with uniform field just imitate Helmholtz coils arrangement, i.e. take two identical permanent magnets with diameters equal or bigger than the distance between their near, attracting poles.  There is still a problem with "vertical" parts of the yoke, they will distort the uniformity of the field within the working area.  Nevertheless, again due to the symmetry, this influence will be equal to zero in vertical symmetry plane.