Synopsis:
In the first 421 square degrees of actual Two-Micron All-Sky Survey (2MASS) data, we have identified 20 objects with spectral types later than M9.5 V as spectroscopically confirmed using the Low Resolution Imaging Spectrograph (LRIS) at the W. M. Keck Observatory. These were found by searching the 2MASS data for objects having no optical counterparts in the POSS-I survey, an IR detection with Ks < 14.5, and colors of either J-Ks > 1.30 (like GD 165B) or J-Ks < 0.4 (like Gl 229B). The total number of ``post-M'' objects now known is 26 after including GD 165B, Gl 229B, Kelu-1, and three cool objects discovered by the DEep Near-Infrared Survey (DENIS). Near-infrared and optical images of one of these discoveries are shown in Figure 1.

How are these objects different from M dwarfs?
Because the TiO and VO bands which dominate the far-optical and near-infrared portions of late-M spectra disappear in these cooler dwarfs, we will define a new spectral class --- spanning roughly 2000-1500K --- where metallic oxides are replaced by metallic hydrides and neutral alkali metals as the major spectroscopic signatures. Figure 2 shows the spectral sequence along with three late M dwarfs. These very cool objects will be assigned to a new spectral class known as ``L''.

Why ``L''?
Of the remaining letters, it's the one closest to ``M'' in the alphabet. See Table 1.

What are these new features in the spectra?
Spectral features are identified in Table 2. Figure 3 shows comparisons of line and band strength changes as a function of temperature for two feature-rich portions of the far optical spectrum.

Why does the sequence end at L8?
All of the 2MASS objects discussed here have 1.30 < J-Ks < 2.10. We have not yet, despite deliberately searching for them, found any dwarfs with J-Ks > 2.10. We believe that dwarfs slightly cooler than those with J-Ks ~ 2.10 have, because of methane formation, much bluer J-Ks colors. Figure 4 shows that methane has not yet appeared at K-band even for our two coolest L dwarfs though it is strong in Gl 229B. However, Figure 3 (rightmost panel) shows that the coolest L dwarf is spectroscopically very similar to Gl 229B in the far optical. We feel that our coolest objects are very close to defining the end of the L sequence and that objects with methane belong in yet another new class, T. To be on the safe side, we truncate the sequence now at L8 to leave space for L9 if needed.

Are any of these brown dwarfs?
Yes. At least six of the 2MASS L dwarfs show the 6708-A lithium doublet at low resolution. See Figure 5. For objects this cool, the presence of lithium proves that they are substellar. Another four objects appear to have lithium lines at the limit of our detectability (~1 Angstrom equivalent width) which if verified means that at least half of the 2MASS L dwarfs are bona fide brown dwarfs.

(Thanks to Ben Oppenheimer and Tom Geballe for the use of their spectra of Gl 229B.)