HomeUpSearchMail
NEW

MegaSound: Sound in Irish neolithic buildings

• Fourknocks I
• Dowth South
• Neolithic dwelling

• Bathroom at home
• Bedroom at home

Tools

Some tips

• When using the RadioShack SPL meter, put it on WEIGHTING C for full frequency response.
• Use ETF in Full Range measurements.
• Give the equipment time to adjust to different temperature and damp environment (to reduce condensation).
• Check noise level with speaker muted. Check if there is constantly 50/60 or 100/120 Hz or other frequencies present (do this with the 3D low frequency waterfall graph). Try to eliminate these grid-power or other related spurious frequencies.
• Check noise level with microphone not connected, but having the cable which would go to microphone in the computer.
• Measure the frequency response of the equipment, by putting the microphone just in front (maximum some 20 cm) of speaker (in free air or in a room with as low as possible reflections) and then check the low frequency response of this system.
• One can type in this frequency response of the measurement system into a mic.cal file, and thus compensating the room measurements. After this is done, don't change any frequency control (like weighting or equalizing) on microphone or amplifier/speaker, otherwise repeat the measurement of the frequency response of the equipment.
This procedure has not been used in the below measurements for Fourknocks I, Dowth South, bathroom and  bedroom. So the power level at the peaks is a result of the room resonance and the output of the speaker and the output of SPL meter at that frequency. So in these cases, one can only say that at a peak a resonance frequency exists, but not if it is the strongest!
• Put speaker as close as possible to a corner, just to excite as many resonance frequencies as possible.

Frequency response of equipment

Frequency response of Ampeg equipment (microphone some 20 cm from speaker)

The Logitech Z-540 performed much better (used in bathroom and iron-age dwelling:

Frequency response of Logitech Z-540 equipment (microphone some 5 cm from speaker)

Fourknocks I

Frequency response (between 20 and 200 Hz) at point K (Low level and door open)

Dowth South

Average of frequency response (between 20 and 200 Hz) at Low level

After the white noise of ETF, a clear tone was heard inside the chamber. It was at least audible for some 0.25 sec and its frequency somewhere around 600 Hz (although I have no musical ears, it could be off with a few 100 Hz).

Bathroom at home

Average of frequency response (between 20 and 200 Hz) at High level

The blue lozenges, purple squares and yellow triangles are the theoretical resonance frequencies resp. axial, tangential and oblique modes.

Bedroom at home

Frequency response (between 20 - 200 Hz) at point A and High level

Considerations

• Human sound production:
• The human voice pitch (f₀: the fundamental frequency of the human vocal cords) is at present days for normal male speaking voice between 110-130 Hz and for normal female speaking voice between 200-230 Hz. Children is around 300 Hz. Singing can produce different pitches of course (higher and lower of some ??? Hz).
Remark: Pitch is sometimes also used just as a synonym to frequency of a sound and sometimes it is related to the perceived frequency of a sound (expressed in [mel] instead of [Hz]). But on this web site it is used only for the fundamental frequency of the vocal cords f₀.
• In former times people were smaller, so it is expected that the pitch would have been higher. Now a days British males are around 174 cm, so compared to European neolithic times (164.2 cm) the pitch could have been some 6 % higher (assuming a linear behavior between pitch and height). So this gives the following pitch frequency ranges:

	Present times Avg. speaking pitch (fo) [Hz]	Neolithic times Avg. speaking pitch  (fo) [Hz]
Male	~120	~127
Female	~215	~228
Child	~300	~320

• Equipment/methodology:
• Looking at the frequency response of my measurement equipment used at Fourknocks I and Dowth South; it has some 32 dB at 56 Hz attenuation. That is quiet a lot. According to a web source, the SPL meter has some 8 dB at 20 Hz  (or around 2 dB at 50 Hz) attenuation. So it looks the speaker/amplifier is not really a sub-woofer system;-).
• The above mentioned frequency response of my Ampeg measurement equipment also causes the peaking around the 100 Hz range. If one would compensate for the measurement equipment, the peaks of the resonance frequencies would be quite flat in the 20 - 200 Hz range.
• For measurements in bathroom and  iron-age dwelling I used a Logitech Z-540 subwoofer system (35 - 200 Hz), this performed much better than the Ampeg system down to 35 Hz (at -35 dB).
• A good source for making a sub woofer can be found here.
• The equipment that Devereux and Jahn ([2002], pers. comm.) used, had a frequency range between 20 - 200 Hz (speaker) and 32-10,000 Hz (microphone).
• The peaks found by Devereux and Jahn ([2002], pers. comm.) were done by listening to the sound. Because the equal-loudness contours of the human ear are quite steep at low frequencies (difference of some 30 dB), real peaks below 100 Hz are perhaps not heard due to this.
• The rooms:
• Looking at the above measurements in rooms, the frequency responses of the measured (badly rebuild) megalithic chamber are not very different from modern house rooms.
• A theoretical sensitivity study using Monte Carlo analysis has been done to determine the chance of an arbitrary room having no axial modes of certain values and the chance of modal coloration according to Gilford. This study revealed the following (using rooms with x, y and z sides ranging randomly between 1.5 and 6 [m] and using some 5500 runs with Monte Carlo analysis):

Pitch range	Chance of no axial modes in pitch range	Chance of coloration  according to Gilford
Now (105-130 Hz)	10%	31%
Neolithic (115-140 Hz)	10%	31%

• A comparison with an iron-age dwelling (with thatched walls and roof) is done. Results will be published soon.
• Some benchmarks:
• When the crossover frequency F₂ is high (like in the bathroom), the theoretical normal modes can be easily seen back in the frequency response graph. For the other rooms an easy one to one mapping of the normal modes and the peaks in the graphs is not that apparent.
• The crossover frequency for Dowth and Fourknocks are resp. around: 100 and 60 Hz. Above these frequencies, normal modes will not be able to explain the measured freq. response, but also the sound ray theory comes into play (angle of incidence equals angle of reflection).
• The one used for the above Monte Carlo analysis is on coloration (quote from Everest [2001], page 351):
"Colorations caused by acoustic anomalies of ... rooms are devastating to speech quality. Gilford states that axial modes spaced approximately 20 Hz or more, or a pair of modes coincident or very close, are frequent sources of colorations. He also states that colorations are likely to be audible when an axial mode coincides with a fundamental (f₀) or first format (F₁) of at least one vowel sound of speech, and are in the region of high-speech energy. Speech colorations below 80 Hz are rare, because there is so little energy in speech in that part of the spectrum. There are essentially no speech colorations above 300 Hz for either male or female voices. Modal colorations are more noticeable in speech than in music."
• Some considerations about the present results:
• It is quite difficult from the above measurements to determine a possible preference (coloration) for male or female usage of the buildings, because only measurements up to 200 Hz have done.
• All the above measured megalithic rooms have rebuild ceilings. It is recommended to do measurements also in as intact as possible chambers like; Cairn L (Loughcrew), Maeshowe (an ideal acoustical setup, which such smooth walls) and Newgrange. This also will provide a better comparison with already measured megalithic rooms by Devereux and Jahn [2002].
• Newgrange seems to be also an ideal Helmholtz resonator, but is it? I have a few queries on this. A Helmholtz resonator must have a defined chamber and a 'pipe'.
Where do you see the two coming together:

• Is it around C1-C21?
  If it is C1-C21, the area of the pipe is not much smaller than the chamber area, in which case the Helmholtz resonance formula is not really valid...
• Or could it be around L17-R15 (O'Kelly [1982], page 99)?
L17-R15 is where the ceiling of the passage is low (again). I think this is the most likely situation for the original building.

• Or is it L11-R10?
  There could be a problem with the lowest ceiling above the L17-L19/R15-R17 which could have an acoustical coupling with the 'pipe (due to possible holes in this roof). In the case, the roof is slowly rising from L11-R10 towards the chamber.
  
• Where does the passage start (after R3-L3)? This because there is a relative large gap due to the roofbox end.
• Furthermore there is K1 nearby the entrance (only 164 cm away and 64 cm high). Would that not change the behavior at such low frequencies?
  
• So it is not really an easy case (for the 'original' situation). In the 'present' case it could even be more difficult due to the many holes in the passage and the modern enclosure. But one could perhaps measure it with a sound source inside the chamber (and frequency between 0.5 to say 10 Hz).
• A more detailed modeling is being investigated.
  
• My idea is that megalithic buildings were rare in neolithic times and because most dwellings were of wood/thatch/open spaced, the totally different acoustic experience could make the megalithic buildings extra special. Acoustic measurements in such an replica of an iron-age dwelling have just been done, the results will be published soon.
• I am not convinced yet that the megalithic chambers were specially build to color the human voice (around 95-120 Hz according to Devereux [2002], page 88). I think though that the special acoustic properties for these rooms (compared to the normal dwellings) could have been used in ceremonies.

Acknowledgments

Disclaimer and Copyright
HomeUpSearchMail