That's a weird dataset. The graph on this page appears to mostly show a progression of record crossing times, not necessarily typical crossing times - with the

That's a weird dataset. The graph on this page appears to mostly show a progression of record crossing times, not necessarily typical crossing times - with the exception that there appears to be one datapoint in the set somewhere around 1912? which shows a higher time than the previous datapoint - at the time the Mauretania held the record from 1909 which would stand until 1929. Given the timing, I wonder whether it's meant to be representative of the journey time expected of the Titanic - perhaps the record speed achieved by a White Star Line Olympic class ship at the time? It's possible that the dataset consists just of record journey times by certain notable ships. It's certainly not all the blue riband records (why are none of the times from 1846 to 1855 plotted?). But then it's not clear what the notable ship that took ~9 days in 1856? was (probably the Persia, but... why is that more notable than the Canada's 8.5 days in 1849?)