where a is the aperture, x is the subject distance and y is the circle of confusion (typically taken as 1/1500th of the field of view in the subject plane).
Then P is the furthest point and S is the nearest point that will appear acceptably sharp in the image. So the depth of field is the distance between P and S.
I believe that this diagram correctly shows the geometrical optics of the situation and that the depth of field can be computed geometrically from this diagram without using lens equations and diagrams of the optics in image space.
The object space diagrams used by Merklinger (that show the rays from a point in the plane of focus) seem to me to be unhelpful to an understanding of depth of field. I'd be interested to know of any good reasons to prefer Merklinger's diagrams to the one above.