A calibration method having the steps of: forming a test image on a sheet; reading the test image on the sheet; and determining correction data to converge data of the test image onto target data, based on information read in the reading step; wherein in cases where in the step of forming the test image a first test image is formed on a face of a transparent sheet as the test image, in the step of reading the test image the first test image is read from a first side of a face to be read which is opposite to the face of the transparent sheet where the first test image is formed.