$\bf{
<2}$

OUR EVALUATION

$\bf{<2.05}$ 
95 
^{ 1} 

SPEC 
$<2.3$ 
95 
^{ 2} 

SPEC 
• • • We do not use the following data for averages, fits, limits, etc. • • • 
$<5.8$ 
95 
^{ 3} 

ASTR 
$<21.7$ 
90 
^{ 4} 

BOLO 
$<5.7$ 
95 
^{ 5} 

ASTR 
$<2.5$ 
95 
^{ 6} 

SPEC 
$<2.8$ 
95 
^{ 7} 

SPEC 
$<4.35$ 
95 
^{ 8} 

SPEC 
$<12.4$ 
95 
^{ 9} 

SPEC 
$<92$ 
95 
^{ 10} 

SPEC 
$15$ ${}^{+32}_{15}$ 



SPEC 
$<19.6$ 
95 


ASTR 
$<7.0$ 
95 
^{ 11} 

SPEC 
$<7.2$ 
95 
^{ 12} 

SPEC 
$<11.7$ 
95 
^{ 13} 

SPEC 
$<13.1$ 
95 
^{ 14} 

SPEC 
$<9.3$ 
95 
^{ 15} 

SPEC 
$<14$ 
95 


ASTR 
$<16$ 



ASTR 
$17\text{ to }40 $ 

^{ 16} 

SPEC 
^{1}
ASEEV 2011 report the analysis of the entire beta endpoint data, taken with the Troitsk integrating electrostatic spectrometer between 1997 and 2002 (some of the earlier runs were rejected), using a windowless gaseous tritium source. The fitted value of ${\mathit m}_{{{\mathit \nu}}}$, based on the method of Feldman and Cousins, is obtained from the upper limit of the fit for ${{\mathit m}^{2}}_{{{\mathit \nu}}}$. Previous analysis problems were resolved by careful monitoring of the tritium gas column density. Supersedes LOBASHEV 1999 and BELESEV 1995 .

^{2}
KRAUS 2005 is a continuation of the work reported in WEINHEIMER 1999 . This result represents the final analysis of data taken from 1997 to 2001. Various sources of systematic uncertainties have been identified and quantified. The background has been reduced compared to the initial running period. A spectral anomaly at the endpoint, reported in LOBASHEV 1999 , was not observed.

^{3}
PAGLIAROLI 2010 is critical of the likelihood method used by LOREDO 2002 .

^{4}
ARNABOLDI 2003A $\mathit et\mathit al.$ report kinematical neutrino mass limit using ${{\mathit \beta}}$decay of ${}^{187}\mathrm {Re}$. Bolometric AgReO$_{4}$ microcalorimeters are used. Mass bound is substantially weaker than those derived from tritium ${{\mathit \beta}}$decays but has different systematic uncertainties.

^{5}
LOREDO 2002 updates LOREDO 1989 .

^{6}
LOBASHEV 1999 report a new measurement which continues the work reported in BELESEV 1995 . This limit depends on phenomenological fit parameters used to derive their best fit to ${{\mathit m}^{2}}_{{{\mathit \nu}}}$, making unambiguous interpretation difficult. See the footnote under ``${{\overline{\mathit \nu}}}~$Mass Squared.''

^{7}
WEINHEIMER 1999 presents two analyses which exclude the spectral anomaly and result in an acceptable ${{\mathit m}^{2}}_{{{\mathit \nu}}}$. We report the most conservative limit, but the other is nearly the same. See the footnote under ``${{\overline{\mathit \nu}}}~$Mass Squared.''

^{8}
BELESEV 1995 (Moscow) use an integral electrostatic spectrometer with adiabatic magnetic collimation and a gaseous tritium sources. A fit to a normal Kurie plot above $18300  18350$ eV (to avoid a lowenergy anomaly) plus a monochromatic line $7  15$ eV below the endpoint yields ${{\mathit m}^{2}}_{{{\mathit \nu}}}$ = $4.1$ $\pm10.9$ eV${}^{2}$, leading to this Bayesian limit.

^{9}
CHING 1995 quotes results previously given by SUN 1993 ; no experimental details are given. A possible explanation for consistently negative values of ${{\mathit m}^{2}}_{{{\mathit \nu}}}$ is given.

^{10}
HIDDEMANN 1995 (Munich) experiment uses atomic tritium embedded in a metaldioxide lattice. Bayesian limit calculated from the weighted mean ${{\mathit m}^{2}}_{{{\mathit \nu}}}$ = $221$ $\pm4244$ eV${}^{2}$ from the two runs listed below.

^{11}
STOEFFL 1995 (LLNL) result is the Bayesian limit obtained from the ${{\mathit m}^{2}}_{{{\mathit \nu}}}$ errors given below but with ${{\mathit m}^{2}}_{{{\mathit \nu}}}$ set equal to 0. The anomalous endpoint accumulation leads to a value of ${{\mathit m}^{2}}_{{{\mathit \nu}}}$ which is negative by more than 5 standard deviations.

^{12}
WEINHEIMER 1993 (Mainz) is a measurement of the endpoint of the tritium $\beta $ spectrum using an electrostatic spectrometer with a magnetic guiding field. The source is molecular tritium frozen onto an aluminum substrate.

^{13}
HOLZSCHUH 1992B (Zurich) result is obtained from the measurement ${{\mathit m}^{2}}_{{{\mathit \nu}}}$ = $24$ $\pm48$ $\pm61$ (1$\sigma $ errors), in eV${}^{2}$, using the PDG prescription for conversion to a limit in ${\mathit m}_{{{\mathit \nu}}}$.

^{14}
KAWAKAMI 1991 (Tokyo) experiment uses tritiumlabeled arachidic acid. This result is the Bayesian limit obtained from the ${{\mathit m}^{2}}_{{{\mathit \nu}}}$ limit with the errors combined in quadrature. This was also done in ROBERTSON 1991 , although the authors report a different procedure.

^{15}
ROBERTSON 1991 (LANL) experiment uses gaseous molecular tritium. The result is in strong disagreement with the earlier claims by the ITEP group [LUBIMOV 1980 , BORIS 1987 ($+~$BORIS 1988 erratum)] that ${\mathit m}_{{{\mathit \nu}}}$ lies between 17 and 40 eV. However, the probability of a positive $\mathit m{}^{2}$ is only 3$\%$ if statistical and systematic error are combined in quadrature.

^{16}
See also comment in BORIS 1987B and erratum in BORIS 1988 .
