10 



the statement with another to the effect that there is always a space 

 reserved for the hairs. 



There are antennas that appear to be wholly scaled ; they are so, 

 certainly, in the sense that there is no special area reserved for hairs. 

 It would seem quite useless for an antenna to be wholly scaled in the 

 sense of having no hairs whatever, as it would then be destroyed as 

 a sense-organ. But it is the case that in Nepticula and other Incom- 

 pletae the scales form a continuous coat, as perfect on one aspect as 

 another. These cases all occur tolerably near the phylogenetic base, 

 whilst the structures are fairly plastic. 



The Gracilariadse (including Lithoco/Ieiis, Ornix^ etc.) have wholly 

 scaled antennae. They have the rather unusual allowance of only 

 one row of scales to a segment, the ruling figure being two rows. 

 Two or more hairs are visible on each segment protruding through 

 the scales. As this family is one of the highest of the Incompletae, 

 the single row of scales is probably an indication that the wholly 

 scaled antenna is not here a direct derivation of the Jugate or Incur- 

 varia antenna, but arose from some form in which the first row of 

 scales disappeared, giving place to hairs, and that the hairs finally hid 

 themselves beneath the scales. We shall see that there are many 

 forms in which the first row of scales tends to disappear in favour of 

 hairs, and some in which it actually does so. I presume that short 

 hairs or rods are present in such an antenna as that of Gracilaria, 

 hidden away beneath the scales, but I have not demonstrated them. 

 There are in some Gracilariadse a few hairs visible as a ring at the base 

 of the scales. 



In Lyonetiadje {Cemiosioma, etc.) the scaling closely resembles that 

 of Gracilaria . 



Some of these antennae, with single complete rings of scales, and 

 no evident hairs, are very beautiful objects, each ring following the 

 last, like a string of flowers (tulips or lilies) threaded one above 

 another. 



I cannot say how it happened that two rows of scales to a segment 

 came to be so dominant a pattern. We have its first appearance in 

 that I have described as belonging to Incurvaria prrelafella, with two 

 circles of scales to a segment, with two rows of hairs alternating. 

 Probably that allowance was determined by the length of an antennal 

 segment in the species that first developed it. 



In Tinea {tapetzella, ^riinella zxidipellionella, etc.) we find this form 

 simplified into one circle of hairs and one circle of scales. The 

 circle of hairs is so wide that the question at once arises, has not the 

 basal row of scales disappeared from a two-ringed antenna ? That 

 this is a correct surmise is proved when we find in Tinea fulvimi- 

 trella and ochraceella portions of the first ring of scales still existing. 



Tinea ochraceella is almost identical with the pellionella form noted 

 above ; but there appear to be two lengths of scales, so that, though 

 there is the appearance of there being only one row, there are really 

 two. There is little or no orientation of the hairs or scales. The 



