Hi all,
in some topics I`ve seen that posters tend to disregard one of the above, according to their opinion of where the truth is.
As I do not wish to start a flame war please allow me to write a few thoughts based on more then 20 yrs experience in research work and teaching at academic level.
The main problem seems to be how far you can generalise accounts and/or tests.
Let`s look at archery as an example. Tests carried out today in order to serve as a good approximation of medieval reality must recreate:
- hardness and tensile strength of armour and arrowhead,
- good spread of realistic striking angles,
- good spread of bow power. Bows of same type had variation in this and the same bow could show variation with, say air humidity.
These few variables already call for lots of test shots if you want to make a prediction within a 95% confidence interval, the minimum acceptable in research. In RL you actually have more variables than that.
OTOH the infamous welsh arrow going through everything...
Well, the door could have been rotten... The arrow could have stricken between the studs in the doors and in the "eye" of mail rings... And so on. What chance is there of such circumstances happening in 95% of shots?
I think that a more humble approach to the validity of witness accounts and test results as a basis for general conclusions may be called for. If I have erred there myself, I humbly apologize to all I may have offended.
Regards, Oldtimer
in some topics I`ve seen that posters tend to disregard one of the above, according to their opinion of where the truth is.
As I do not wish to start a flame war please allow me to write a few thoughts based on more then 20 yrs experience in research work and teaching at academic level.
The main problem seems to be how far you can generalise accounts and/or tests.
Let`s look at archery as an example. Tests carried out today in order to serve as a good approximation of medieval reality must recreate:
- hardness and tensile strength of armour and arrowhead,
- good spread of realistic striking angles,
- good spread of bow power. Bows of same type had variation in this and the same bow could show variation with, say air humidity.
These few variables already call for lots of test shots if you want to make a prediction within a 95% confidence interval, the minimum acceptable in research. In RL you actually have more variables than that.
OTOH the infamous welsh arrow going through everything...
Well, the door could have been rotten... The arrow could have stricken between the studs in the doors and in the "eye" of mail rings... And so on. What chance is there of such circumstances happening in 95% of shots?
I think that a more humble approach to the validity of witness accounts and test results as a basis for general conclusions may be called for. If I have erred there myself, I humbly apologize to all I may have offended.
Regards, Oldtimer