Atomic Data Revisions for Transitions Relevant to Observations of Interstellar, Circumgalactic, and Intergalactic Matter

Throughout this paper, we focus on the electric dipole (E1) transitions. The absorption line corresponds to the transition between the lower level l and the upper level u, with level energies ${E}_{{\rm{l}}}$ and ${E}_{{\rm{u}}}$, respectively. The statistical weights ($g=2J+1$, J being the total angular momentum of the state) of the lower and upper levels are denoted ${g}_{{\rm{l}}}$ and ${g}_{{\rm{u}}}$. In terms of the energy levels, the vacuum wavelength ${\lambda }_{\mathrm{Ritz}}$ of the transition is

$$\begin{eqnarray}&&{\lambda }_{\mathrm{Ritz}}=\mathrm{hc}/({E}_{{\rm{u}}}-{E}_{{\rm{l}}}),\end{eqnarray} \tag{ 1 }$$

where h is Planck's constant, and c is the speed of light. The Einstein transition probability (in s⁻¹) for spontaneous emission is denoted by ${A}_{\mathrm{ul}}$. The dimensionless absorption oscillator strength ${f}_{\mathrm{lu}}$ is related to the E1 transition probability ${A}_{\mathrm{ul}}$ by

$$\begin{eqnarray}&&{f}_{\mathrm{lu}}=1.49919\times {10}^{-16}{\lambda }_{\mathrm{vac}}^{2}{g}_{{\rm{u}}}{A}_{\mathrm{ul}}/{g}_{{\rm{l}}}\end{eqnarray} \tag{ 2 }$$

where ${\lambda }_{\mathrm{vac}}$ is in Å. The oscillator strengths compiled in this paper were obtained from either experimental techniques or theoretical calculations. In cases where oscillator strengths in both the length and velocity forms are available, we tabulate the length-form data because these are commonly used and more reliable.

In Table 1, we present the data for each ion species separately. These species and their ground levels are given in table subheaders. Table 1 lists the following information for each E1 line of interest.

• 1.  
  Nuclear charge Z.
• 2.  
  Ion spectroscopic notation.
• 3.  
  Lower level.
• 4.  
  Upper level.
• 5.  
  Vacuum rest wavelength ${\lambda }_{\mathrm{vac}}$ in Å (where available).
• 6.  
  Ritz wavelength ${\lambda }_{\mathrm{Ritz}}$ in Å.
• 7.  
  Lower level statistical weight ${g}_{{\rm{l}}}.$
• 8.  
  Upper level statistical weight ${g}_{{\rm{u}}}.$
• 9.  
  Absorption oscillator strength f.
• 10.  
  Logarithm of weighted oscillator strength, $\mathrm{log}({gf}).$
• 11.  
  Accuracy grade (uncertainty) of the oscillator strength f according to the NIST atomic spectra database (Kramida et al. 2015).
• 12.  
  f-value data source.

Table 1.  Line Identifications, Observed Wavelengths ${\lambda }_{\mathrm{vac}}$ (Å), Ritz Wavelengths ${\lambda }_{\mathrm{Ritz}}$ (Å), Absorption Oscillator Strengths f, and Their Accuracy Grade for Key Transitions

Z	Ion	Lower level	Upper level	${\lambda }_{\mathrm{vac}}$	${\lambda }_{\mathrm{Ritz}}$	${g}_{{\rm{l}}}$	${g}_{{\rm{u}}}$	f	$\mathrm{log}({gf})$	Grade	Source
6	C i	2s22p2 3P0	2s22p3s 3Po1	1656.928	1656.929	1	3	1.43E−1	−0.845	A	FF06
6	C i	2s22p2 3P0	2s2p3 3Do1	1560.310	1560.309	1	3	7.16E−2	−1.145	A	FF06
6	C i	2s22p2 3P0	2s2p3 3Po1	⋯	1328.834	1	3	5.80E−2	−1.236	B	FF06
6	C i	2s22p2 3P0	2s22p4s 3Po1	⋯	1280.135	1	3	2.61E−2	−1.583	B+	FF06
6	C i	2s22p2 3P0	2s22p3d 3Do1	1277.245	1277.245	1	3	9.22E−2	−1.035	A	FF06
6	C i	2s22p2 3P0	2s22p5d 3Do1	1157.910	1157.910	1	3	2.12E−2	−1.674	⋯	ZF02
6	C ii	2s22p 2Po1/2	2s2p2 2D3/2	1334.532	1334.532	2	4	1.29E−1	−0.589	⋯	FFT04
6	C ii	2s22p 2Po1/2	2s2p2 2S1/2	1036.337	1036.337	2	2	1.19E−1	−0.624	⋯	FFT04

Note. An estimated accuracy grade is listed for each oscillator strength where available, indicated by a code letter as given below: AAA $\leqslant \,0.3 \% ;$ $0.3 \% \,\lt$ AA $\leqslant \,1 \% ;$ $1 \% \,\lt$ A+ $\leqslant \,2 \% ;$ $2 \% \,\lt$ A $\leqslant \,3 \% ;$ $3 \% \,\lt$ B+ $\leqslant \,7 \% ;$ $7 \% \,\lt$ B $\leqslant \,10 \% ;$ $10 \% \,\lt$ C+ $\leqslant \,18 \% ;$ $18 \% \,\lt$ C $\leqslant \,25 \% ;$ $25 \% \,\lt$ D+ $\leqslant \,40 \% ;$ $40 \% \,\lt$ D $\leqslant \,50 \% ;$ E $\gt \,50 \%$.

Only a portion of this table is shown here to demonstrate its form and content. A machine-readable version of the full table is available.

Download table as:  DataTypeset image

We wish to be as consistent as possible in selecting the lines presented in Table 1. Hence we need to explain a selection procedure for the tabulated lines. The following basic rules were followed for selecting data sources in our tables or providing any additional information.

• 1.  
  For the species range, we consider elements with $6\leqslant Z\leqslant 30$ and several other elements observed in the ISM, such as Ga, Ge, Kr, and Pb. Usually, we present data for a few of the lowest ionization stages. The selected ions have been observed in the ISM, CGM, and/or IGM.
• 2.  
  We tabulate lines with wavelengths ${\lambda }_{\mathrm{vac}}\gt 911.753\,\mathring{\rm{A}}$. These are the vacuum wavelengths for all lines.
• 3.  
  We give priority to the observed wavelength, which we call ${\lambda }_{\mathrm{vac}}$ over the Ritz wavelength ${\lambda }_{\mathrm{Ritz}}$. As a rule, the line wavelength source is the NIST database (Kramida et al. 2015). We have represented the Ritz wavelengths from the NIST database to three decimal places.
• 4.  
  We tabulate absorption lines originating from the ground level only. We do not tabulate lines originating from the excited levels of the ground configuration or the ground term even if their energies are just a few tenths of cm⁻¹, e.g., the C i, C ii, and N ii ions.
• 5.  
  We tabulate only those lines that have $f\geqslant 0.001$. We present no more than three significant figures for the f-values because we suppose that is enough to reflect their real accuracy.
• 6.  
  Usually, we tabulate f-values from the newest sources giving priority to the experimental data over the theoretical values. In the cases where new data are not significantly different from the older data, we choose to rely on the older data preferring the most advanced theoretical methods for data production. Those special cases will be mentioned in Section 3.

The accuracy grades of the tabulated oscillator strengths f were either derived from the original data sources or we tabulated the grades given in the NIST database by Kramida et al. (2015). For this reason, some of the lines do not have accuracy grades. In some cases, we were able to confidently assign an accuracy grade through careful comparative analysis. These instances and their justification are specifically documented within Section 3.

For neutral carbon and the species C ii and C iii, we tabulate theoretical data from the calculations of Froese Fischer and co-workers published in Froese Fischer (2006), Zatsarinny & Froese Fischer (2002), and Froese Fischer & Tachiev (2004). These are very reliable data sources providing highly accurate radiative transition data. For two lines in the lithium-like C iv, which is outside the scope of the above-mentioned papers, we adopt data from Yan et al. (1998) with an accuracy grade of A. We performed an additional check for these two lines and calculated transition data using our own codes utilizing the Hartree–Fock (HF) and quasirelativistic (QR) approaches on a very extensive configuration-interaction (CI) wavefunction basis recently described by Kisielius et al. (2015). Our results confirm the high accuracy of data from Yan et al. (1998); our calculated f-values agree within a few tenths of a percent.

Oscillator strengths are taken from Froese Fischer & Tachiev (2004) for the species N i, N ii, and N iii. The NIST database assigns them accuracy grades no worse than B. There is just one line in our investigated wavelength range in the N iv ion. It originates from the excited level 1s²2s2p ³P₁. However, it is rather weak because it constitutes a spin-changing E1 transition. For the lithium-like ion N v, we adopt data from Peach et al. (1988). For the N v lines, we performed an accuracy check because these data are coming from non-relativistic calculations. Our results, both in the length and velocity forms of the E1 transition operator, agree between themselves within 2% and do not deviate more than 0.5% from the results of Peach et al. (1988). This confirms a high-accuracy grade of A assigned to oscillator strengths of the tabulated absorption lines in the N v ion.

For the O i ion, data from Froese Fischer & Tachiev (2004) are adopted for the lines connecting the excited states 2s²2p³3s, 3d, and 4s. For other lines not covered by Froese Fischer & Tachiev (2004), we use data from Butler & Zeippen (1991) and Hibbert et al. (1991). The latter is the same source used by the NIST database, and the later investigation of Tayal (2009) just confirms the reliability and high accuracy of their data. Transition data from Tayal (2009) agree with oscillator strengths f of Hibbert et al. (1991) within 10%. We also tabulate oscillator strengths for two 2s–2p transition lines in Li-like O vi determined by Peach et al. (1988). Likewise to the situation described in Section 3.2, we have performed our own calculation and can confirm a high-accuracy grade assigned to these data by the NIST team.

The oscillator strength f-values are taken from Froese Fischer et al. (2006) for the lines 3s–np, (n = 3, 4), and we retain the same accuracy grades as given by NIST. For the line 3s–5p, we adopt the oscillator strength value determined by applying the same MCHF approach B-spline method with non-orthogonal radial orbitals by Froese Fischer (2002). It was proved for neutral carbon by Zatsarinny & Froese Fischer (2002) that such an approach produces reliable results for the high-nl lines.

For the lines representing resonance transitions from the ground 3s² state to the excited 3s3p and 3s4p states of the Mg i ion, we use data from Froese Fischer et al. (2006) with accuracy grades of A and B+. For the transition to the 3s5p levels, more recent oscillator strength f data produced using a B-splines method within MCHF by Zatsarinny et al. (2009) are available, whereas for the lines representing transitions to highly excited 3snp levels (6 ≤ n ≤ 8), we adopt data of Chang & Tang (1990) because more recent data do not exist.

Oscillator strength values for Mg ii are taken from Froese Fischer et al. (2006). The accuracy of the data for the 3s–3p lines has a very high A+ grade. Meanwhile, though the lines 3s–np ($n\gt 3$) are in our investigated wavelength range (${\lambda }_{\mathrm{vac}}\gt 911.753$), only the transition to the 2p⁶5p ²P${}_{3/2}^{{\rm{o}}}$ level has $f\geqslant 0.001$ satisfying our selection criteria. Other lines fall short of that criteria, though their fine-structure level unresolved oscillator strengths are within that range.

We tabulate oscillator strength f-values for three ions, Al i, Al ii, and Al iii. Here for the lines connecting the ground state with the lower $n=3,4$ levels, the theoretical data of Froese Fischer et al. (2006) are adopted. They have accuracy grades of B+ and A+. For the higher nl levels, we list theoretical data from Mendoza et al. (1995), which were determined by non-relativistic R-matrix calculations. Furthermore, their data for fine-structure levels were derived from the multiplet values assuming a pure LS-coupling. Those data usually have rather poor accuracy grades, thus further data improvement can rectify this situation. For three lines of Al i, we adopt experimental data of Vujnović et al. (2002) and Davidson et al. (1990) and assign an accuracy grade of B.

For the low-lying states of the 3s²3p4s, 3s3p³, 3s²3p3d odd-parity configurations of Si i, we list f-values adopted from Froese Fischer et al. (2006). As we have mentioned before, these theoretical data are of high quality and reliability. It is necessary to admit here that these values differ noticeably from the previous oscillator strength values tabulated by Morton (2003). The case of the lines involving higher levels is more complicated. Some lines falling into our wavelength selection region are absent from the NIST database, though their corresponding level energies are presented there, e.g., the lines at ${\lambda }_{\mathrm{Ritz}}=1568.617\,\mathring{\rm{A}}$, ${\lambda }_{\mathrm{Ritz}}=1763.661\,\mathring{\rm{A}}$, and ${\lambda }_{\mathrm{Ritz}}=1873.103\,\mathring{\rm{A}}$. The oscillator strength value for the line at ${\lambda }_{\mathrm{Ritz}}=1589.173\,\mathring{\rm{A}}$ is derived from the non-relativistic R-matrix calculations of collision data for several Si-like ions of Nahar & Pradhan (1993), which we consider to be less accurate compared to the pure atomic structure calculations or the experimental results. The E1 transition data for the latter line are listed in Froese Fischer et al. (2006), though their $\mathrm{log}({gf})$ value of −2.16 differs noticeably from the experimental log (gf) = −2.57 value from Smith et al. (1987). Additional experimental data for higher nl levels comes from the measurements of Smith et al. (1987) and from a critical compilation of Kelleher & Podobedova (2008).

Bautista et al. (2009) have produced a benchmark data set for the Si ii ion by utilizing several theoretical approaches and experimental data to determine reliable recommended absorption oscillator strengths for the levels of the 3s3p², 3s²3d, and 3s²4s configurations. For the higher level of 3s²4d at 989.873 Å, we list the experimental f-value of Curtis & Smith (1974), which is close to the theoretical value f = 0.1849 from Charro & Martin (2000), which also serves as a source for the 3s²5s level data. For the ions Si iii and Si iv, we list oscillator strengths from Froese Fischer et al. (2006).

For the lines connecting the ground state 3s²3p³ ${}^{4}{{\rm{S}}}_{3/2}^{{\rm{O}}}$ with the levels of ${}^{4}{\rm{P}}$ term of the excited configuration 3s²3p²(${}^{3}{\rm{P}}$)4s of P i, we adopt oscillator strengths from Froese Fischer et al. (2006). Unfortunately, that work does not list data for the transitions to the 3s²3p²(${}^{3}{\rm{P}}$)3d ${}^{4}{\rm{P}}$ levels. Therefore, for these levels, we resort to older theoretical data from Fawcett (1986). The f-values for the lines at $\lambda \lambda 1679.695,1674.591,1671.680$ given by Froese Fischer et al. (2006) differ significantly, at least by two orders of magnitude from the previous data of Fawcett (1986). Since the relative intensities for these three 3s–3p lines given by the NIST database are similar to those of 3p–4s lines at $\lambda \lambda 1787.656,1782.838,1774.951$, we believe the data from Fawcett (1986) given its D accuracy rating can be utilized for neutral phosphorus.

For the P ii and P iii lines, we list data from Froese Fischer et al. (2006). Five more lines below 1153 Å that are unlisted in the NIST database are entered in Table 1 for P ii. These lines originate from the absorption transitions from the ground level to the levels 3p4s and 3p3d. We have derived their Ritz wavelengths using level energies listed by Kramida et al. (2015). The same process was performed for three P iii lines below 999 Å. The source of the transition data chosen for the magnesium-like P iv and sodium-like P v is Froese Fischer et al. (2006).

The main source of the data for the neutral sulfur lines connecting the ground level 3s²3p^{4 3}P₂ with the levels of the excited 3d, 4d, 4s, 5s, and 6s configurations are the theoretical results of Deb & Hibbert (2008). As in Deb & Hibbert (2006), the data from Deb & Hibbert (2008) agree very well with most lines from the results of the B-spline calculations by Zatsarinny & Bartschat (2006). They are also close to the data produced by Froese Fischer et al. (2006). Thus one has to be assured of high quality and reliability of the listed f-values. We include the line at $\lambda =1474.5715\,\mathring{\rm{A}}$ even though its f-value is lower than our selection criterion $f\leqslant 0.001$ as other theoretical predictions put it above this criterion. The data from higher configurations with the valence 5d, 6d, 7s, and 8s electrons are taken from Biémont et al. (1998). That set of f-values was deduced from a combination of laser lifetime measurements and theoretical branching ratios. For the lines with $\lambda \lt 1241\,\mathring{\rm{A}}$ that connect the ground state with the upper levels located above the first ionization limit at 83,559.1 cm⁻¹, we choose oscillator strengths from Deb & Hibbert (2008), which have an accuracy grade of C.

For the S ii lines, we list a data set from Kisielius et al. (2014). The authors concluded that their data are in good agreement with other theoretical data sets of Froese Fischer et al. (2006) and Tayal & Zatsarinny (2010). For the S iii, S iv, and S vi lines, we list data from Froese Fischer et al. (2006).

The data for the Cl i lines are taken from the recent theoretical calculations of Oliver & Hibbert (2013). Their data accuracy grade is C+. For the two lines in Cl ii and three lines of Cl iii, we assume that the experimental data from Schectman et al. (2005) are the best source. The accuracy of their f-values is within 5% (or an accuracy grade of B+). We list an oscillator strength from Froese Fischer et al. (2006) for a single line of Cl iv.

For the lines of Ar i and Ar ii, we recommend using oscillator strengths from Froese Fischer et al. (2006). One can safely assign the C+ accuracy grade to these data.

For the spectra of neutral K i, we list four lines corresponding to the resonance 4s–4p and 4s–5p transitions in the valence shell. The lines to the higher np levels are relatively weak and do not fall into our desired f-value range. The parameters of lines caused by the 4s–4p transitions are determined by Wang et al. (1997). Their data agree very well with other high-accuracy measurements. They evaluate the accuracy of f-values as 0.2% (AAA accuracy grade). Data for the 4s–5p lines are from the measurements of Shabanova & Khlyustalov (1985). Although their accuracy grade is unlisted, by comparing these data with the results of elaborated relativistic calculations of Migdalek & Kim (1998), we confidently assign them a grade of A.

The data for the λ4227.92 line are taken from the high-accuracy photoassociative measurements of Zinner et al. (2000), which produces reduced uncertainties compared to previous measurements. Their data have confirmed previous level-crossing measurements of Kluge & Sauter (1974). Recent theoretical values of Froese Fischer & Tachiev (2003) are also close (within error bars) to the above experimental data. The data for other resonance lines of neutral calcium are measured by Parkinson et al. (1976). They used the hook method to determine the gf-values of the 4s²–4snp absorption lines. These data for most lines agree within 5% with other measurements, e.g., Ostrovskii & Penkin (1961), Shabanova (1963), and Mitchell (1975). The λ2722.450 line is the only exception, and here the differences are much higher. Even though this line has an f-value below our cut-off level, we list it in our table. The list for Ca i contains lines up to 3p⁶4s13p, while the transitions for higher Rydberg levels fall below our cut-off criterion of $f\geqslant 0.001$.

For the Ca ii ion lines, we recommend recently determined oscillator strengths from Safronova & Safronova (2011). Their calculations used a high-precision relativistic method, where all single and double, and partially triple excitations of Dirac–Fock wavefunctions are included to all orders of perturbation theory. The authors estimate the relative uncertainties of their calculated oscillator strengths to be 0.9%.

The majority of the data for the Ti i lines come from the measurements of Lawler et al. (2013). They have used the previously measured radiative lifetimes combined with the branching fractions to yield absolute oscillator strengths. The data agree closely with the NIST data; therefore, we consider it appropriate to assign them respective accuracy grades, which are no worse than B+. Some lines are missing in Lawler et al. (2013). For those lines, we recommend the high-quality experimental data from Blackwell-Whitehead et al. (2006) or data from the earlier experiment of Smith & Kühne (1978).

For the Ti ii ion lines, we list data from the recent paper of Lundberg et al. (2016). The authors measured radiative level lifetimes and used the pseudo-relativistic HF method to determine oscillator strengths and theoretical lifetimes for the measured levels. Their data agree within 10% with the earlier theoretical data of Ruczkowski et al. (2014) and those from Pickering et al. (2001b). For the Ti iii transition data, we adopt the theoretical results of Raassen & Uylings (1997). They performed an orthogonal operator calculation for the electric dipole transition integrals by means of the multiconfiguration Dirac–Fock method including core polarization.

For the lines connecting the ground state of Cr i with the n = 4 levels, the data chosen were produced by Sobeck et al. (2007). The authors used branching fraction measurements from Fourier transform spectra in conjunction with radiative lifetimes to determine the transition probabilities. These data are assigned an accuracy grade of B. For the transitions to the 3d⁵5p,6p levels, we list data from Martin et al. (1988), which are given accuracy grades of C.

For the three lines above 2050 Å in Cr ii, the recommended tabulated data are from the experimental work of Nilsson et al. (2006). The accuracy grade for their data is B+. For other lines of Cr ii, we tabulate theoretical data from Raassen & Uylings (1998), where oscillator strengths have been determined using the orthogonal operator technique. Five listed lines of Cr iii are taken from Kurucz (2016).

For the manganese lines, we adopt the experimental oscillator strength values from Blackwell-Whitehead et al. (2005, 2011). Their accuracy grade is B+ for the lines with $\lambda \gt 4000\,\mathring{\rm{A}}$, and C+ for the lines below 3000 Å. Unfortunately, oscillator strengths for most Mn i lines listed in our table were not measured in the above experiments. For these lines, we tabulate older data from Martin et al. (1988) or data from Kurucz (2016).

New data for some selected Mn ii lines in Table 1 are listed from Den Hartog et al. (2011) who experimentally measure radiative lifetimes and branching fractions to derive transition probabilities and oscillator strengths. Other data come either from similar experimental work of Kling & Griesmann (2000) or from the configuration-interaction calculations of Toner & Hibbert (2005).

Lines for three iron ions are listed in Table 1. For lines with $\lambda \gt 2913$ Å, our oscillator strength source is Blackwell et al. (1979), who measured absorption oscillator strengths and reported an accuracy of 0.5%. NIST assigns an accuracy grade of A for most of those lines. For the lines below 2913 Å, our data come from the measurements of O'Brian et al. (1991). They employed time-resolved laser-induced fluorescence to measure radiative lifetimes and derived oscillator strengths by measuring branching fractions. Their accuracy grades are slightly worse compared to those of Blackwell et al. (1979). For the remaining lines below 2260 Å, we list the data of Banfield & Huber (1973), who used the hook method to determine oscillator strengths in Fe i. Although the accuracy grade of their data is not as high, these oscillator strengths are substantially reliable.

Our main source for the Fe ii lines above 2000 Å is the experimental data from Bergeson et al. (1996). The authors have measured branching ratios with a Fourier transform spectrometer and with a high-resolution grating spectrometer. The resulting measurements were used to derive transition probabilities for 56 lines. Another group of listed lines was theoretically studied by Raassen & Uylings (1998). Their accuracy grade is lower. These two sources are complemented by experimental oscillator strengths from Pickering et al. (2001a) and the critical compilation of Fuhr & Wiese (2006). For the Fe iii ion line at 1122.5 Å, we tabulate the theoretical oscillator strength from Deb & Hibbert (2009).

Oscillator strengths for all but one of the Co i lines are taken from the recent measurements of Lawler et al. (2015). They derived oscillator strengths from experimental branching fractions combined with radiative lifetimes from laser-induced fluorescence measurements. The NIST database assigns accuracy grades of B or B+ for their data. A rather weak line at λ2436 is taken from Cardon et al. (1982), where absolute oscillator strengths were determined using the hook method. Data for the Co ii lines come from the calculations using the orthogonal operator technique by Raassen et al. (1998) and the measurements of Mullman et al. (1998a, 1998b). In general, the data for Co ii have relatively low accuracy grades. For the lines of Co iii, we tabulate oscillator strengths from the Kurucz online data list (Kurucz 2016).

For the Ni i lines, we list oscillator strengths from new measurements of Wood et al. (2014), where f-values were determined by combining measured branching fractions and radiative lifetimes. Most of these data are assigned high-accuracy grades (from B+ to B). Missing data are covered by the measured oscillator strengths obtained by the hook and absorption methods from Huber & Sandeman (1980). Their data accuracy evaluation is lower compared to that of Wood et al. (2014). NIST lists only a single line for transitions of Ni ii. We tabulated oscillator strengths from the recent CI calculation of Cassidy et al. (2016). The authors state that it is difficult to provide a measure of the uncertainties for the large-scale CI calculations, which would cover all transitions. Nevertheless, they consider that their data are the best currently available and meet the accuracy demands for astrophysical applications. We do not assign any accuracy grades to these data, though they can be considered as having an accuracy grade of B.

The source of oscillator strength data for the resonance 4s–4p lines in Cu i is a critical compilation of oscillator strengths for neutral ions by Doidge (1995). The data for the transitions 3d–4p and 4s–5p are adopted from Hannaford & McDonald (1978). The oscillator strengths were determined from atomic absorption measurements of the radiation emitted from a copper hollow-cathode lamp. The oscillator strengths for the remaining lines 4s–np $(n\geqslant 6)$ comes from the calculation of Çelik et al. (2015).

Three lines listed for Cu ii originate from the transitions involving the ground level 3p⁶3d^{10 1}S₀ and the levels with J = 1 from the configuration 3d⁹4p. Their data source is either the theoretical data from Donnelly et al. (1999) or from the beam-foil experimental data of Brown et al. (2009), which were given an accuracy grade of B.

The oscillator strength for the 4s^{2 1}S₀–4s4p¹P^o₁ line of Zn i is taken from the Hanle-effect experiments of Kowalski & Träger (1976), which have confirmed the earlier level-crossing technique results of Lurio et al. (1964). Later beam-foil spectroscopy measurements of Martinson et al. (1979) give very similar f-values ($f=1.55\pm 0.08$). These measurements are consistent with recent theoretical data, e.g., Froese Fischer & Zatsarinny (2007) and Głowacki & Migdałek (2006). For the $\lambda 1589$ line, we tabulate the oscillator strength from the line list of Kurucz (2016). The NIST database lists a rather strong line at $\lambda 1109.1$ with no transition identification or Ritz wavelength. Analyzing the level list, we can assume that this line originates from the transition 3d–4p with ${\lambda }_{\mathrm{Ritz}}=1108.316\,\mathring{\rm{A}}$, though this wavelength noticeably differs from the measured one. This can be explained by the fact that the level 3d⁹4s²4p ³P${}_{1}^{{\rm{o}}}$ lies in the continuum, above the ionization limit.

Oscillator strengths for Zn ii are taken from the quasirelativistic calculations of Kisielius et al. (2015). Their data are consistent and agree with other theoretical data within a range of 10%.

Tabulated data for neutral gallium lines are from a new compilation of Shirai et al. (2007) with an unlisted line for the 4p_1/2–7d_3/2 transition data taken from the relativistic many-body perturbation theory calculation of Safronova et al. (2006). These two data sources tabulate oscillator strengths differing by approximately 10% for most of the listed lines. For the λ1414 line of Ga ii, we tabulate an oscillator strength from the low-energy beam-foil measurements of Andersen et al. (1979), which have 8% measurement errors (i.e., an accuracy grade of B). Later theoretical data confirm the reliability of the listed experimental data, see, e.g., McElroy & Hibbert (2005), Jönsson et al. (2006), and Chi & Chou (2014). Oscillator strength values for two lines of the Ga iii ion are from the compilation of Shirai et al. (2007).

There are no new reliable original oscillator strength data sources for the germanium species following Morton (2003), except for the compilation of Fuhr & Wiese (2005). For the Ge i lines, we tabulate experimental oscillator strengths of Li et al. (1999). Their experiment measured natural level radiative lifetimes by employing time-resolved UV laser-induced fluorescence from a laser-produced plasma and determined branching fractions by an inductively coupled plasma emission spectrometry technique. The derived oscillator strengths have accuracy grades of B or B+. For the Ge ii lines, we list theoretical f-values from Biémont et al. (1998). For the Ge iii ion, we list two 4s²–4s4p transition lines. The ³P^o₁ spin-forbidden line oscillator strength is tabulated from the empirical predictions of Curtis (1992). The resonance line f-value for the ¹P^o₁ level was experimentally determined by Andersen et al. (1979) utilizing beam-foil spectroscopy. The oscillator strengths for Ge iv are adopted from the beam-foil measurements of Pinnington et al. (1981).

For the Kr i lines we tabulate experimental oscillator strengths from Chan et al. (1992). They measured photoabsorption f-values using the dipole ($e,e$) method. The oscillator strengths have accuracy grades of B or B+. The reliability of their data was confirmed by recent B-spline calculations of Zatsarinny & Bartschat (2009). The oscillator strengths for Kr ii were derived from the selective pulsed monochromatized synchrotron radiation experiment of Lauer et al. (1999), where lifetimes of the 4s4p^{6 2}S${}_{1/2}$ states were measured. For the Kr vi line, we list data from MCHF relativistic calculation of Pagan et al. (1996), where adjustments were made to the electrostatic parameters in order to improve theoretical level energies and gf-values.

We list data for the Pb ii lines. The oscillator strength for the $\lambda 1682$ line is adopted from Quinet et al. (2007), where transition probabilities were calculated in a relativistic multiconfiguration Dirac–Fock approach and the accuracy of the results was assessed by a new experimental determination of the radiative lifetime for the 7s${}_{1/2}$ level. Other listed data are from Safronova et al. (2005). The authors obtained radiative transition data using relativistic many-body perturbation theory. The remaining data for the 10s${}_{1/2}$ and 9d${}_{3/2}$ levels are adopted from the ab initio relativistic Hartree–Fock calculations of Colón & Alonso-Medina (2001).