19 Association of Research Libraries Research Library Issues 295 — 2018 As we’d expect, we see a significant drop as the individuals in the first cohort age into the next, until 2010–2015, the period in which delayed retirements became a factor. The large percentage of individuals aged 65 and over in 2015 is an aberration, and it begs the question as to what will happen to this group in the 2020 data. This happens to be a question we can answer with reasonable certainty. The key lies in isolating those individuals aged 70 and over across the data sets. Doing so reveals that very few ARL professionals work past age 70, about 1% in every data set since 1986. It is certainly possible that, in 2020, those who were in the 65+ cohort in 2015 might suddenly prove to be the first to work past 70 in large numbers, but this is highly unlikely. The 2020 data will almost certainly reflect the retirement of at least 8% of the ARL population from the 2015 65+ cohort alone, the equivalent of over 800 individuals. (See Figure 3.) For perspective, 8% retirement in this cohort would be 2.5 times the number retiring between 2005 and 2010, and almost 5 times the number retiring between 2000 and 2005. Figure 3 Given the imminent surge of retirements from the oldest age cohort, it is worth asking whether the vacancies that result will be filled on a one-to-one basis. Money for positions is precious and often fraught politically, but here again, the data are clear in support of replacement