Astronomical CCDs are fabricated as a two-dimensional array of tiny pixels (picture elements) on a thin wafer of semiconductor, usually silicon; there may be up to several thousand rows and columns of pixels. When light or other radiation is directed onto this array, each pixel responds to the photons falling on it by producing electrons. Electric charge thus accumulates in each pixel in proportion to the amount of incident radiation. After an exposure, these packets of charge are shifted out of the array and the accumulated charge in each pixel is measured, row by row. The values are digitized and stored in a computer, and may be used to form an image on a computer screen or may be further manipulated or analyzed. There is a direct relationship between the intensity of the recorded image and the original exposure, hence the linear response. Noise is, however, introduced as the charges are moved out of the CCD, amplified, digitized, and stored in the computer, thus placing a lower limit on the signal that can be accurately recorded; this readout noise can be reduced by cooling the CCD.