https://web.archive.org/web/20070402230356/http://www2.gol.c...
Each lunar month is 29 or 30 days long, so if we miscalculate the month then the error will be also 29/30 days, back or forth. However, the graph in the article shows that almost all dates fall in (modern calendar) April.
So, if the blossoms peaked at 4/20 and we miscalculate by one month, we'll arrive at 3/21 or 3/22, which is too early to be likely. Ditto in the other direction. So, even in the worst case, we can safely assume that most of the dates shown are correct (or maybe one day off, due to disagreement over when the new moon started).
In addition, if they were meticulous enough to record cherry blossoms, I'd assume that the Japanese Imperial court would have written how many months were in each year ...
OTOH, looking at the data, events are dispersed over more than a month (range 124 - 86 = 38 days), so it is possible an earlier date was pushed later, where it fits better - one cannot reliably deduce month from year + day of month.
I found a tool that attempts to back-convert Japanese dates into Gregorian dates. The tool is called NengoCalc. There's a page (in German) here http://www.yukikurete.de/nengo_chronology.htm that talks about the challenges of the historical Japanese calendar. Here's a quote from the google-translate version of that page:
"In Japan, the Gregorian calendar was introduced in the early Meiji period. The following applies: ５ 明治 12 月 3 日 corresponds to 1.1.1873. Before that, the bound lunar calendar from China was used. With this the new moon falls on the first of the month. There are months with 29/30 days. From time to time intercalendar months were added, which were named after the previous month. In the case of Japan, it is particularly difficult because this did not happen regularly and therefore there was no continuous time base and consequently no continuous year counting.
Instead, you have the system of annual currencies, era names, 年号. These can be changed at any time, e.g. September 1615: 慶 長 20 年 -> 元 和 １ 年. The first currency of the year was 大化 (645). Because of the lunar calendar, the annual currency year can have a time difference of 4-8 weeks to the Gregorian calendar, so you need appropriate tables for conversion. So it can be that at the end of an annual currency year in the Gregorian calendar you are already in the following year. At this point it should also be pointed out that in Japanese materials this fact is often not taken into account and simply incorrect information is given. Since 1868 there has only been one currency per reign. If the annual currency changes, the year is named after the new currency. In addition, this is also used as a posthumous name for the emperor."
The tool NengoCalc, uses three different tables that have been developed for converting the dates and shows the discrepancies between those tables, and tries to use the "most correct" one, but verifying the dates is impossible.
All of this is to say that making claims about "this is happening on the earliest day ever" is a good headline, but scientifically unverifiable because there's no way to tie the dates in Japanese historical records to an astronomical calendar.
Anyway, feel free to look for a valid possible earlier day.
Every calendar has specific days with full moons because that’s a physical event that takes place regularly. It doesn’t matter when you’re adding a lunar month, those full moons will always fall on the same days, thus full moon + 7 days for example can only ever be 7 days from one of those full moons on any calendar.
Years are more complicated but in 864, Japan adopted one of the Chinese calendars in which a year was defined as 365.2446 which is quite close to the actual 365.2425 days and thus worked fine for long periods.
However, the 44 minutes adjustment needs a leap day every ~2.6 years or it obviously drifts out of sync. Adding a full month is pointless because it doesn’t change the offset.
it's not difficult to recalculate the dates. it's been done for most other anniversaries for historical events.
Eventually you go back far enough and this breaks down, but we are talking recent history here.
They should be transparent about it and they weren't, and a lot of people here seem to be willing to just trust them and their methodology, but I can't figure out what's scientific or transparent about that. Isn't peer review supposed to catch and correct these types of issues before publication? And if not, shouldn't they be transparent enough in sharing their methodology that the concerns raised by amateur idiots like myself would be easily addressed, or at least accounted for?
Anyway, you have clearly already decided the issue in your own mind and are looking to justify it. So, no I am not going to waste my time providing evidence you can look up if you’re willing to consider the possibility you’re wrong then dare to do it yourself.
You’re effectively complaining they didn’t show their work when using dates from the Roman etc calendar. Historians really care about getting dates correct and have done extensive work to line up each historical calendar where possible.
This is the paper. It did describe the challenges of inferring the full-bloom date from old literary work. But the calendar conversion is not one of the challenges.
This is a list of material and conversion tables composed by the National Diet Library on the old calendars. There are clearly well-established method and researches about this. This would be the most basic problem for anyone studying history in Japan. It may be not straightforward (for you), but it certainly would be too generic and elementary for other historians to read in a paper like this.
Isn't the Japanese calendar Lunisolar and borrowed from China? (from the wikipedia link you posted above)
Since the paper doesn’t say this, I assume you’re one of the researchers or associated with them? Do you mind sharing some details about how it’s taken into consideration?
here's one of them : https://link.springer.com/article/10.1007/s00484-009-0272-x
The historical Japanese calendar suffered from drift away from seasonal events because the months had either 29 or 30 days in them, but the actual lunar calendar is 29.53... days long. This was a known problem, and they would occasionally add a 13th month to the year to compensate, but this was only done when either the Emperor died, or when the Emperor said to do it (usually to commemorate some significant battle). Both of those situations means that the leap months were not added in a systematic way, so trying to back-calculate dates has errors. Several attempts have been made over the last 75 years by scholars to create conversion tables for this, but they all have errors and can be inaccurate up to 4 weeks.
Have you thought of composing your thoughts in this thread into article and posting it on Medium/somewhere?
I would be happy to give it a proofread if you write it up. Contact details in DM.
I don't understand why so many people are dismissing the calendar issue and just assuming that "it's accounted for" or "it's not relevant" or "it's trivial" or "it's just noise". Where's the transparency? They've shared their dataset, but not the method used to determine the dates? That looks like a huge oversight and so many people here seem to be accepting the premise of the claim and giving the benefit of the doubt to the researchers. What's scientific about that?