HARVARD UNIVERSITY 
The Biological laboratories 
16 DIVINITY AVENUE 
CAMBRIDGE 38, MASSACHUSETTS 
10 July, 1964 
Dear Martin: 
I have finally finished your Sonagraphs of Callicebus * 
it was a bigger -job than I had expected it to be. 
The principal problem was with certain sounds like the 
grunt which arc both very low-pitched and very brief. Griffin's toy 
will go down to quite low frequency levels, but it does this by changmgg 
sax*x sppeds of some of its devices, and so has to compensate in time- 
in short, to analyse a very low sound one is stuck with a condensed time 
scale in which the length of the Sonagraph paper increases from the 
usual 2.4 secs to 8 secs. A very brief sound then becomes almost a 
simple point - you see its frequency , but no other characteristics. 
(Oddly, by the converse of the above, very high-pitched sounds are ana- 
lysed on an expanded time scale, and the various trills etc about which 
you were so worried are really ragnificent - you can see all sorts of 
beautiful detail.) 
On the whole, after a good deal of fiddling, I think we've 
come out quite well. Most of the sounds were not difficult to analyse; 
wherever 'here was any difficulty for any reason, I tried to do a 
number of different types of analysis - or, rather, I tried to use a 
variety of different frequency and time scales. I'll send some inter- 
pretation along with the analyses and tape (they 'LL go registered, for 
which I'll have to wait until Monday). 
I've burned up about $8 worth of paper on the $ob, but we 
can compensate for this if I'm doing further work for you. The postage 
1(11 try to get the MCZ to pay - maybe by sending it in Mayr4f name. 
I don't know if we'll run into a problem with brief, low- 
pitched sounds on many of your species. I've checked out the liter- 
ature I have from the bonagraph Co. on the model I'm going to order, 
and, so far as I can be sure, it looks likely to be a considerable 
improvement. It's normal analysis will be 2.4 secs of sound at from 
85 up to 8,000 cps, but it can apparently expand this same up to 4kc in- 
stead or give a scale up to 850cps(and maybe expand this to 425cps) 
with a special pug-in filter I can obtain for it. These 4 basic choices 
would apparently all use the 2.4 sec time scale, and, if so, are prefer- 
rable to Griffin's machine. I did find his 2,500 and 1,500 scales 
useful and won't be able to match them on ny machine. All his scales are 
lines®, however, and I can apparently use a logarithmic scale on mine - 
nice because it better approximates what an animal hears and so should be 
more meaningful (I don't know *hat the £ range of this log scale is). 
Finally, for extremely brief calls there is a method by which I shall be 
able f .o expand the time scale by 3 tines, although 1 doubt if this will 
be necessary. 
