Crying Dragon

I went to Nikko Toshogu Shrine on my recent school trip. Underneath the "Crying Dragon" in the Yakushido Hall, a temple staff member used wooden clappers to make a sound, and I was able to hear the dragon's voice.

That's good to hear. Did they explain why the dragon roars?

Yeah. They explained it was because of the repeated reflections between the ceiling and the floor, but thinking about it now, it's kind of strange why it only makes noise where the dragon is.

I have detailed information about the "Crying Dragon," so let me explain it to you. The dragon painting was originally created by Kano Eishin Yasunobu, but the Yakushido temple where the Crying Dragon is located burned down in a fire in 1961, and in 1968, the dragon painting was restored by artist Katayama Nanpu.

Was it known for the dragon's roar even before the fire?

That's right. So, in order to reproduce that sound, they conducted model experiments to confirm its effectiveness, and apparently restored not only the image but also the sound.

Wow. How did they conduct that model experiment?

A paper published in 1965 [1] * describes the results of measurements taken with the actual structure in 1953 before the model experiment, stating: "The number of repeated reflected waves is 29 times/second, it takes 2.5 seconds for the sound to attenuate by 60 dB, and because of the 6 cm curvature (the ceiling is a two-dimensional concave surface), the sound of clapping hands in the center resonates for a long time due to the back-and-forth reflection between the ceiling and the floor." As shown in this figure (Figure 1), observations made with an oscilloscope at the time revealed a continuous series of reflected wave pulses, which represents the phenomenon described as a dragon roaring. The paper states that the model experiment (1/4 scale) was conducted to observe the size of the curvature and the differences in waveforms depending on the position of the sound source, and to clarify what was necessary to reproduce the phenomenon.

But even without a curve, if there are parallel surfaces, wouldn't it result in back-and-forth reflection?

That's a good question. If the ceiling is a concave surface, even if sound enters the ceiling at an angle, it will reflect back in the direction of incidence, so the reflected waves will concentrate at the location of the sound source (receiving point). However, if the ceiling is flat, sound that enters directly from above will undergo multiple reflections parallel to the floor in place, but if it enters at an angle, the reflection path will move outward by that angle, away from the center. Therefore, if the ceiling is a concave surface, the number of repeated reflected waves will be greater, and the reflection time will be longer.

But strangely enough, the sound of the wooden clappers sounded like a "beep" to me. It wasn't a repeating "clang" sound, so I was a little surprised.

That's another mystery of the Crying Dragon. I explained earlier why the sound is repeated and creates an echo, but there's another reason why it sounds different from the original. First, when humans hear short-duration sounds (like a popping sound: called a pulse) in succession, if the interval between sounds is shorter than 50 ms (0.05 seconds), the sounds don't sound separate and are heard as one sound. Conversely, if the interval between sounds is longer than 50 ms, they sound separate. If the original sound wave is a pulse, it becomes a flutter echo, which sounds like a flapping sound.

Sound travels about 340 meters per second, right? So, if two sounds travel 17 meters apart in 0.05 seconds, they'll be heard separately, but if they're any further apart, they'll merge into one sound, is that correct?

To be precise, the point at which the sound is heard is called the receiving point. A time difference of 50 ms or a distance difference of 17 m between two sounds reaching that receiving point determines whether they are perceived as two separate sounds or as a single sound. Therefore, considering the length of the reflection paths, the individual reflected sounds of the Nikko Crying Dragon should not be heard separately. (*Note 1)
I have a sample of what it sounds like with a continuous sound, so listen to it.
This is a sample of sound produced by applying a periodic time variation to a single frequency (a pure tone) of 1000 Hz, starting with a slow variation and gradually increasing the speed of the variation. Even with a continuous sound, if its period exceeds 50 ms (20 Hz), we can still capture that period.

The frequency here indicates how many times it changes per second.

That's right. If you listen to the notes in order, starting from 1 Hz which occurs once per second, did you notice anything?

Just like my dad said, the sounds started to become indistinguishable from around 20 Hz, and at frequencies above 50 Hz, they sounded like some kind of muddy noise.

The reason the sound of the wooden clappers under the painting of the crying dragon sounded like a beep instead of the original "clang" is because of this. When temporal fluctuations occur with the same period, humans perceive the sound as roughest when the frequency of the period (modulation frequency) is around 70 Hz. You said the sound felt muddy, but in other words, you meant that the sound sounded rough, gritty, and rough.

There was something strange about the "Crying Dragon" in Nikko, but now I understand why.