Temperatures along the Tokyu Toyoko Railway Line between Shibuya and Yokohama recorded atevery 15-second intervals by a thermistor sensor placed between the doors on the side of the first car of a train traveling the 25km of the line. One of the features of the distributions was a rise in temperature during the train's stops at stations. To identify the cause of that phenomenon, the rises in temperature were compared for trains passing through and those stopping at the same stations. The data show that the rise in temperature for trains stopping is as twice large as for those passing through. In the case of moving trains, errors in reading the thermometer caused by time lag should betaken into account, calculated according to the following conditions: The temperature around the platform changed in steps as model distributions. A typical station on the Toyoko Railway Line was chosen as a model station, Velocity distribution was that at the head of a train passing through the model station. After the errors were accounted for, the temperature rise for slowing became the same as that for stopping. As the time lag of the thermistor is small under this condition, the rise in temperature at the time of the tram's stopping at the station may have other momentary causes like the heat created by brake friction. In reality the temperature is not judged to change in steps owing to the advection currents of air. The temperature measured by a moving thermistor is not that at the point of observation but a spatial moving average temperature. The data excluding those after the stopping of a train and those recorded in 14sec after the train starts are useful. If the time interval for routine observations is within 8sec or less, the heat islands created by around urban stations can be observed.
Many Japanese workers have recognized that landslides can be classified into two types; in type (1) the speed is higher, the duration is shorter and the movement rarely reactivates; in type (2) the speed is lover, the duration is longer and the movement continues intermittently for a long period. The former is named here A-type and the latter, B-type, Previous classifications Which were based on experience are qualitative and descriptive. Therefore, it is necessary to quantitatively classify these two types. The maximum speed and the duration of movements recorded at 120 landslides on natural slope in Japan were quantitatively investigated on the basis of the existing literature. The movement of the A-type landslide is divided into (1) pre-failure movement, (2) main movement and (3) post-failure movement., as shown in Fig. 1 as an example. The distance and speed of the pro failure andpost-failure movements are so small as to be negligible, compared with those of the main movement. Therefore, the speed and the duration for landslides of this type are estimated from the data on the main movement. For the B-type landslide, active and inactive periods alternately repeat (Fig. 2). The duration and the maximum speed are estimated from one active period. The histograms of duration and maximum speed constructed using the existing data are shown in Fig. 3. The duration histogram (Fig. 3-A) has a hi-modal distribution, which can be divided into two groups at the bottom of trough of the histogram, i.e., 101.5 (_??_32) hours. Dual-peak distribution is found in the speed histogram (Fig. 3-B); a trough between the two peaks is located at 10-2 to 10-1m/min. Fig. 3-B also shows that (1) landslides with a duration of less than 32 hours have a higher speed than 10-2m/min, and (2) the landslides with a duration of more than 32 hours have a lower speed than 10-1 m/min. The above discussion leads to the following classification. A landslide having higher speed than 10-1.5m/min and shorter duration than 32 hours is named the “rapid-type”, and one having lover speed than 10-1.5m/min and longer duration than 32 hours is named the “slow-type”