We discuss the feasibility of detecting the progenitors of normal elli
ptical galaxies (PGs) undergoing their first massive starbursts at lar
ge redshifts using narrow-band imaging in the near infrared. Four stro
ng emission lines are plausible tracers of such objects, viz. [O II] l
ambda3727, Hbeta lambda4861, [O III] lambda5007, and Halpha lambda6563
, spanning the redshift range z approximately 1-5.5 in the standard JH
K bands. The expected line fluxes are in the range F approximately 10(
-16+/-1.5) erg/cm2/s, depending on the star-formation history, emissio
n line, amount of dust present, redshift, and cosmology. The brighter
end of this flux range is already within the reach of existing technol
ogy. We estimate the expected surface density of protogalaxies on the
sky and several other related quantities. We report on the results of
a pilot project done at Palomar to search in the K band for [O II] lam
bda3727 emission line galaxies in the fields of three z > 4 quasars, u
sing a narrow-band (DELTAz=0.013) imaging technique. The total area su
rveyed is 0.72 arcmin2, with a limiting line flux (90% confidence limi
t) in the range F approximately (4-10)X10(-16) erb/cm2/s. The total co
moving volume covered is V approximately 100 Mpc3, down to limiting re
st frame line luminosities in the range L approximately (0.3-30) X 10(
43) erg/s, or implied unobscured star-formation rates in the range SFR
approximately (30-15 000) M./yr, for a Friedman cosmology with H-0=75
km/s/Mpc, OMEGA0=0.2, and LAMBDA0=0. No viable candidate protogalaxie
s were found, though this was not entirely unexpected given our limite
d field size and line flux limits, but the limits achieved are an impr
ovement over previously published surveys. Forthcoming experiments sho
uld be able to extend these limits by two or more orders of magnitude
in comoving volume coverage and one or two orders of magnitude in limi
ting flux, and could detect young ellipticals if our calculations are
representative of their emission-line properties.