science
warning! science content follows. let's say the thermometer tells me it's 72 degrees outside my house. but to be useful in a scientific sense you also need to know the error. ie it's 72 degrees plus or minus 2 degrees. now suppose there's a satellite with an infrared detector over my house. it can also measure the temperature at my house. but since it's so far away its error bar is larger. let's say it gets a value of 70 degrees plus or minus 3 degrees. now combine the results and get 71 degrees plus or minus 1 degree. okay 71 degrees no problem but wait just a cotton picking minute. how can the error of the combined measurements be smaller than either error alone? heh. glad you asked. the errors tend to cancel out making the average more accurate than either individual method. this works best when the two methods are independent. like say thermometers and satellites.