What one must remember is that in any scheme of Devanagari vertical metrics, unlike Latin, there is not and cannot be a prescriptive model that tells us exactly where a part of every letter must come in contact with a given metrics line, the way the Latin’s x-height, caps height, ascender, or descender do. While the headline and baseline do form a fixed vertical structure, even the lines denoting the highest and lowest vowel signs are not an indication of which particular vowel and other signs reach them — they help us demarcate the vertical boundaries of a design. Things get murkier still for the remaining intermediary vertical metrics. There is no absolute rule, for instance, about precisely where the knot of the न should fall in the context of the intermediary metrics between the headline and baseline and what else should match this vertical position. It varies from design to design. One must see the metrics, barring headline, and baseline as gradations in space that can be used to describe letterforms and their overall proportions, rather than as strict guidelines in which letterforms and their anatomy must forcibly fit.
Further to his 2009 paper, Anatomy of Devanagari Typefaces, Girish Dalvi has created a scheme for Devanagari letter parts and anatomy, which has been published online at Devanagari Search Tool. In his paper, he points out that a previous scheme by Mahendra Patel is specific to a particular style of typeface; Dalvi tries to rectify that. For example, he not only defines parts like “knot,” but also offers options for how it could be represented: open, closed, or filled, much the same way a serif can be bracketed, wedge, hairline, etc. Several terms, such as contrast, axis, terminal, and counter have been borrowed from Latin type anatomy, which makes the scheme easy to use in multiscript settings.