Before the canoes were launched, a man fastened two upright forked sticks on each side of one, near the middle.  About three and a half feet astern of these a cross-piece was laid on the bottom of the craft.  To this was attached two shorter forked sticks.  Between each pair of upright forked sticks was placed another cross-piece, thus forming two horizontal bars, parallel to each other, one only a few inches from the bottom of the boat and the other about a foot and a half above the gunwales.  Next, four slabs of Caripari wood of varying thickness, about three feet long and eight inches wide, were suspended from these horizontal bars, so as to hang length-wise of the canoe and at an angle of forty-five degrees.  Each pair of slabs was perforated by a longitudinal slit and they were joined firmly at their extremities by finely carved and richly painted end-pieces.

The operator strikes the slabs with a wooden mallet or hammer, the head of which is wrapped with an inch layer of caoutchouc and then with a cover of thick tapir-skin.  Each section of the wooden slabs gives forth a different note when struck, a penetrating, xylophonic, tone but devoid of the disagreeably metallic, disharmonic bysounds of that instrument.  The slabs of wood were suspended by means of thin fibre-cords from the crosspieces, and in this manner all absorption by the adjacent material was done away with.

By means of many different combinations of the four notes obtained which, as far as I could ascertain, were Do—­Re—­Mi—­Fa, the operator was able to send any message to a person who understood this code.  The operator seized one mallet with each hand and gave the thickest section, the Do slat, a blow, followed by a blow with the left hand mallet on the Re slat; a blow on the Mi slat and on the Fa slat followed in quick succession.  These four notes, given rapidly and repeated several times, represented the tuning up of the “wireless,” calculated to catch the attention of the operator at the maloca up-creek.  The sound was very powerful, but rather pleasant, and made the still forest resound with a musical echo.  He repeated this tuning process several times, but received no answer and we proceeded for a mile.  Then we stopped and signalled again.  Very faintly came a reply from some invisible source.  I learned afterwards that at this time we were at least five miles from the answering station.  As soon as communication was thus established the first message was sent through the air, and it was a moment of extreme suspense for me when the powerful notes vibrated through the depth of the forest.  I shall never forget this message, not only because it was ethnographically interesting, but because so much of my happiness depended upon a favourable reply.  I made the operator repeat it for my benefit when we later returned to our village, and I learned it by heart by whistling it.  When printed it looks like this: