Interesting chart at Very Small Array on "the slow death of the instrumental." I think there's some truth to that claim - but what I really think the chart reflects is the dominance of rock'n'roll on the charts - the best-seller charts, that is. (Even though rock is no longer a dominant chart genre, its successor genres in popularity are also rarely instrumental: R&B, hip-hop, even pop-country.) Since '63 or '64, when the Beatles arrived in American and pretty much ushered out pre-rock forms from the singles charts, there are basically two anomalous years for instrumentals: one in the mid-sixties (the design of the chart makes determining an exact year a bit difficult) and one in the early eighties.
Remove those anomalous years, and you have a pretty steady percentage of instrumental hits in post-Beatles years - it looks like under 5%. Note that since there are 52 weeks in a year, each week's charts account for a little less than 2% of the year's total. And that means that a single song that hits number one for, say, six weeks can, all by itself, take up more than 10%. (At least I assume this is how the chart is done - but the alternate method, whereby each song counts only once, would still give longer-charting songs higher weight - because there would then be fewer songs to make up that percentage. However, that method wouldn't differentiate between long-charting songs and short-charting songs: if I were doing the chart, I'd definitely give each week its own "space.")
My guess is that those two anomalous years can be attributed to one, two, maybe three instrumental hits.
The single most striking stat here, though, is that it seems 1996 all but eliminated male vocalists from the charts. Bizarre. (Also, if the graph were rearranged, so the green area were on top to make its line more visible, it might also serve to illustrate "The Decline of the Male-Female Duet." Theme song: "Don't You Want Me" by the Human League...)