Research on the historical grammar of Iranian languages can indulge in data spanning over 3000 years and over a vast region. The field has also been fortunate that a good share of these data were available early enough to be studied by pioneers by eminent scholars such as Christian Bartholomae, Heinrich Hübschmann, Wilhelm Geiger and Paul Horn. While their works are now more than 100 years old, most of them have not been replaced by a more recent works. One reason for this surely is the fact that these pioneers applied the rigorous method of comparative linguistics, by which, for instance, Hübschmann was able to establish that there were two languages, not one, in the material that until then was considered summarily as Pahlavi – viz. Middle Persian and Parthian, as became clear with the discovery of the Manichean manuscripts in Eastern Turkestan – much in the same way as he also established that Armenian is a separate branch of Indo-European rather than a sub-branch of Iranian as had been thought until then. For reasons of lack of data, much of the work by the early scholars had to apply what can be called the “black box method” (a terminology that I borrow from Eichner 1992). Essentially, one thus finds in the pioneering works statements of the type “Old Iranian (or: Proto-Iranian, or: Proto-Indo-European) X yields Middle Persian / New Persian / Balochi / Pashto Y”. This approach, useful as it is for practical purposes, does not take into account the relative chronology of the changes concerned. (To be sure, the pioneering works do occasionally note that change A needs to have happened before change B.) Today, entire corpora of Middle and New Iranian languages have become available, but some traditional methods that the pioneers had no choice but to apply have lingered on; handbooks and articles still perpetuating the “black box” method. In this talk, I will present examples for the “glass box method”, which tries to determine the relative chronology of changes leading from one stage of a language to another. This approach is obviously more precise, but, as I intend to show, also highlights problems with the changes hitherto assumed which are not quite visible in a black box perspective. This state of affairs becomes particularly clear when dealing with a period during which an important number of changes must have happened. One such period is the time between the latest Old Persian inscriptions and the earliest Middle Persian texts. It is perhaps not by accident that the period during which the essential changes between Old and Middle Persian happened is not attested directly. Armenian data combined with the glass box method help to achieve, I think, a coherent chronology of changes, or at least one that accounts for the data, which is not the case when applying the black box method.