Every adopted radiant-point should be based on at least five paths, unless the conditions are special, and these must show a very definite centre, and present family resemblances. It is often possible to detect a good centre from very few paths, when the radiant is low on the horizon, or when it occupies an isolated position.
In recording meteors the details of their appearances should also be appended to the paths. Foreshortened and crooked courses, fluctuations of brightness, halting motion, spark-trains, phosphorescent streaks, broken streaks, and other features must be invariably noted when observed, as likely to assist in fully comprehending these bodies. A streak will sometimes brighten up perceptibly after the head has died out.
One of the principal aims of future observers should be to ascertain the visible duration of meteor-showers, and the displacement or fixed position of the radiants during the period of their continuance. The Perseids seem to endure for forty-six nights (July 8-August 22) while the radiant moves from 3° +49° to 76° +57°. The Lyrids also exhibit a shifting radiant, and it is highly probable some other showers are to be included in the same category. In investigating these, the observations of single nights should be kept separate, and the radiant determined from each set of paths. The positions when compared will then exhibit the rate and direction of the displacement. As to radiants which are apparently stationary[46] during long intervals, these should be closely observed. Are the centres of radiation, as successively determined, identical, allowing for the slight errors of observation? Are they continuously in operation, or intermittent? Meteors with motions in declination and near their radiants will be specially valuable in settling these questions, and if observed at more than one station will possess great significance. If it can be proved that a radiant is fixed and continuous during a few weeks, there can be no reason why it may not be stationary for a much more lengthy interval, unless the circumstances are exceptional.
Though I have pointed out the urgency of noting the directions and durations of meteors, there are other features in such observations that must not be disregarded. If the paths are being recorded for the particular purpose of getting duplicate observations and calculating the heights, then it is desirable to note the beginning-and end-points of the flights as exactly as possible, for unless this is done the combined paths will show great discordances. Those who have acquired a familiar knowledge of the constellations will, however, experience little trouble in insuring accuracy in these records.