A、The one which looks along earthquake faults. B、The one which relies on phenomena. C、The one which is based on a formula calcul

Now, you’ve been reading articles about the tremendous damage done to life and property by earthquakes. That’s why seismologists have been working so hard to develop methods of earthquake prediction. We can now predict earthquake fairly well, but the predictions only locate potential areas of danger. They don’t predict the specific time and location at which an earthquake is likely to occur. Today I want to introduce to you three prediction models that have been developed. The first prediction model looks along earthquake faults, those cracks in the earth’s crust, to find places where the fault has shown little or no earthquake activity for a long time. This theory hypothesizes that such places are due for a major shock. The second model relies on phenomena— like ground tilt. Using long tubes containing water, observers noted that ground tilt tended to occur before major earthquakes. That led them to correctly predict the big Haicheng quake of 1975— the first successful earthquake prediction scientists have ever made. A million people were evacuated from that Chinese city before the earthquake struck. Unfortunately, this method hasn’t worked consistently, so we can’t say it’s been perfected. The third model is based on the theory that major earthquakes closely follow a series of minor ones. Starting with the measurements and timing of the smaller quakes— a complex formula calculates the “times of increased probability” for much larger quakes. Right now, this method, like the first method, cannot predict specific times and places, but that may change as it is further developed. For the moment, none of these models can predict with reasonable levels of confidence.
16. What does the speaker say about the earthquake prediction?
17. What places are more likely to have a major earthquake?
18. Which prediction model helped scientists predict the big Haicheng quake of 1975?
19. What is the third prediction model based on?