The James Clerk Maxwell Telescope has been used to obtain submillimeter and millimeter continuum photometry of a sample of 30 IRAS sources previously studied in molecular lines and centimeter radio continuum .
All the sources have IRAS colours typical of very young stellar objects ( YSOs ) and are associated with dense gas .
In spite of their high luminosities ( L \lower 2.58 pt \hbox { $ \sim$ } \kern - 7.75 pt \raise 2.795 pt \hbox { $ > $ } 10 ^ { 4 } L _ { \odot } ) , only ten of these sources are also associated with a radio counterpart .
In 17 cases we could identify a clear peak of millimeter emission associated with the IRAS source , while in 9 sources the millimeter emission was either extended or faint and a clear peak could not be identified ; upper limits were found in 4 cases only .

The submm/mm observations allow us to make a more accurate estimate of the source luminosities , typically of the order of 10 ^ { 4 } L _ { \odot } .
Using simple greybody fitting model to the observed spectral energy distribution , we derive global properties of the circumstellar dust associated with the detected sources .
We find that the dust temperature varies from 24 K to 45 K , while the exponent of the dust emissivity vs frequency power-law spans a range 1.56 < \beta < 2.38 , characteristic of silicate dust ; total circumstellar masses range up to more than 500 M _ { \odot } .

We present a detailed analysis of the sources associated with millimeter peaks , but without radio emission .
In particular , we find that for sources with comparable luminosities , the total column densities derived from the dust masses do not distinguish between sources with and without radio counterpart .
We interpret this result as an indication that dust does not play a dominant role in inhibiting the formation of the H ii region .
We examine several scenarios for their origin in terms of newborn ZAMS stars and although most of these ( e.g .
optically thick H ii regions , dust extinction of Lyman photons , clusters instead of single sources ) fail to explain the observations , we can not exclude that these sources are young stars already on the ZAMS with modest residual accretion that quenches the expansion of the H ii region , thus explaining the lack of radio emission in these bright sources .
Finally , we consider the possibility that the IRAS sources are high-mass pre-ZAMS ( or pre-H-burning ) objects deriving most of the emitted luminosity from accretion .